Ep 115: The Writing Moderation Mess & Using AI Speed Up Report Writing!

1
00:00:00,040 --> 00:00:02,160
Hello everyone, and welcome back
to another episode of Teach

2
00:00:02,160 --> 00:00:02,840
Sleep.
Repeat.

3
00:00:02,840 --> 00:00:04,640
My name is.
Dylan, and my name is Hayden.

4
00:00:04,640 --> 00:00:07,120
Why, Hayden, have you spent all
morning hating the Black Eyed

5
00:00:07,120 --> 00:00:09,120
Peas?
I haven't, but you've been

6
00:00:09,160 --> 00:00:13,600
singing the same thing.
Oh man, oh man, oh man, man.

7
00:00:13,600 --> 00:00:18,040
God, it's, it's a tune.
It's not, though it takes.

8
00:00:18,040 --> 00:00:20,600
Me back anyone around our age?
I'm 31.

9
00:00:20,600 --> 00:00:24,840
You're 4732.
Actually, that's the same

10
00:00:24,880 --> 00:00:26,800
difference.
Everyone around our age there

11
00:00:26,800 --> 00:00:29,280
was that boom when we're like
teenagers and they I remember

12
00:00:29,280 --> 00:00:31,520
what album it was, but it was
like the album of like the the

13
00:00:31,640 --> 00:00:34,960
computerised Queen, the end.
Oh, we know it was good.

14
00:00:35,000 --> 00:00:37,280
I was a huge banger.
Base band was good, but it was

15
00:00:37,280 --> 00:00:38,960
just it was full of banger off
the banger.

16
00:00:38,960 --> 00:00:41,360
What was the other one that was
like boom, pow, yes, boom, Boom

17
00:00:41,360 --> 00:00:43,200
Pow.
I guess when you look at it,

18
00:00:43,200 --> 00:00:46,240
it's it might not be the most
intricately written song.

19
00:00:46,760 --> 00:00:49,720
Yeah, yeah.
Listen, as like a a big Black

20
00:00:49,720 --> 00:00:51,520
Eyed Peas fan before that, you
know, Monkey Business and

21
00:00:51,520 --> 00:00:52,840
Elephant.
The two hours before that,

22
00:00:52,840 --> 00:00:55,120
obviously I go.
But when this came out, it was

23
00:00:55,120 --> 00:00:56,200
different.
I was like, this is not their

24
00:00:56,200 --> 00:00:57,240
style, you know, when you're a
kid.

25
00:00:57,240 --> 00:00:59,600
And it's like, how dare they.
How dare they experiment with

26
00:00:59,600 --> 00:01:00,360
their genre.
Yeah.

27
00:01:00,640 --> 00:01:02,720
And so I was very annoyed.
But they don't produce.

28
00:01:02,720 --> 00:01:04,200
I don't.
Remember ever being a child and

29
00:01:04,200 --> 00:01:05,960
thinking how dare they
experiment with their genre?

30
00:01:05,960 --> 00:01:07,400
I didn't.
Think that?

31
00:01:07,400 --> 00:01:08,840
Well, we live very different
lives, Dylan.

32
00:01:08,960 --> 00:01:11,640
You know you're out partying to
Boom Boom, Pow and other.

33
00:01:11,640 --> 00:01:13,880
Song pow.
Those something, something man

34
00:01:13,880 --> 00:01:16,000
Pow.
Yeah, that's those people trying

35
00:01:16,000 --> 00:01:17,760
to be Power I.
Think that's how it went, yeah.

36
00:01:17,920 --> 00:01:20,040
Something like that's
indistinguishable.

37
00:01:20,040 --> 00:01:21,520
Can you?
Imagine now if we got writing,

38
00:01:22,160 --> 00:01:25,200
we, our children in school,
produced an independent bear.

39
00:01:25,200 --> 00:01:27,360
Writing that was like that.
We'd be like, what is?

40
00:01:27,360 --> 00:01:28,920
Where's the front of the verbal
here?

41
00:01:29,120 --> 00:01:30,880
But Sir, you said we're going to
write songs today we need.

42
00:01:31,080 --> 00:01:33,520
To have the determiner in the
right place please.

43
00:01:34,640 --> 00:01:36,720
What is Boom Boom power if not
on a matter here?

44
00:01:39,160 --> 00:01:41,160
Take that off the writing list.
You've done that.

45
00:01:41,440 --> 00:01:43,280
You've evidenced that we've.
Got writing moderation coming

46
00:01:43,280 --> 00:01:43,880
up?
Come on, Black Eyed.

47
00:01:43,880 --> 00:01:45,280
Do you know, do you know what
you Speaking of writing

48
00:01:45,280 --> 00:01:46,360
moderation?
I have to tell you about this

49
00:01:46,360 --> 00:01:48,280
because I forgot to mention the
other day to you, right?

50
00:01:48,280 --> 00:01:50,800
I, I saw someone online and this
it ground my gears a little bit

51
00:01:50,800 --> 00:01:52,960
because I was a bit like, oh,
how are we still at this point

52
00:01:52,960 --> 00:01:55,640
where people think this, right?
Someone was saying in a year six

53
00:01:55,640 --> 00:01:59,280
group that they got writing
moderation coming up and they

54
00:01:59,280 --> 00:02:02,280
were like what what's what,
what's a good way to get year

55
00:02:02,280 --> 00:02:06,120
5-6 spelling words in in their
writing so that I can take off

56
00:02:06,120 --> 00:02:09,800
that they can spell.
Because one of the like famously

57
00:02:09,800 --> 00:02:11,720
with with writing moderation,
there's this, there's the

58
00:02:11,720 --> 00:02:13,680
objective of using the national
curriculum words.

59
00:02:13,680 --> 00:02:15,440
Yeah.
And.

60
00:02:15,880 --> 00:02:17,760
So having yeah, any, so anyone
who doesn't know what that is,

61
00:02:17,760 --> 00:02:21,440
it's a list of like 40 to 50
words, right, for the F500.

62
00:02:21,800 --> 00:02:22,520
Wow.
OK.

63
00:02:23,120 --> 00:02:25,320
And, and it's, and it is, it is
it in the curriculum that you've

64
00:02:25,320 --> 00:02:27,080
got to use it just for anyone
because people might not be

65
00:02:27,080 --> 00:02:28,880
aware of this, especially they
have newer teachers and lower

66
00:02:28,880 --> 00:02:31,200
down in the school.
What you're saying is there's a

67
00:02:31,200 --> 00:02:34,120
list from the government of
words that they should learn and

68
00:02:34,120 --> 00:02:35,720
is it to spell or is it to use
in writing?

69
00:02:35,760 --> 00:02:37,200
I think it's a bit of both.
It's very vague.

70
00:02:37,200 --> 00:02:38,920
That's the point.
And people have over the years

71
00:02:38,920 --> 00:02:41,360
we've decided ourselves
obviously as a culture that was

72
00:02:41,360 --> 00:02:43,920
like, OK, well then to prove, to
prove that they can meet that

73
00:02:43,920 --> 00:02:46,240
objective of using, if I explain
words, we must see lots of

74
00:02:46,240 --> 00:02:47,480
evidence of them in their
writing.

75
00:02:47,480 --> 00:02:49,680
We're going to look specifically
for those words and it's just

76
00:02:49,680 --> 00:02:51,320
not true.
It's just not true.

77
00:02:51,320 --> 00:02:53,440
I've been to enough writing
moderations now where everyone's

78
00:02:53,440 --> 00:02:54,680
like, no, that's rubbish.
What you talking?

79
00:02:54,680 --> 00:02:57,720
About, is it not true?
That is what you're saying that

80
00:02:57,720 --> 00:02:58,720
we don't need to use them,
right?

81
00:02:58,720 --> 00:03:01,480
That's what you're saying.
Is that actually the case or is

82
00:03:01,480 --> 00:03:03,240
it classic?
What always seems to happen,

83
00:03:03,240 --> 00:03:05,760
education, is there'll be some
pockets of people doing

84
00:03:05,760 --> 00:03:09,000
moderation where that is seen
and looked for, and therefore

85
00:03:09,000 --> 00:03:11,160
actually those people do need to
do it because you'll be marked

86
00:03:11,160 --> 00:03:12,520
down where it's just
inconsistent.

87
00:03:12,520 --> 00:03:14,880
I don't know.
My gut feeling is honestly just

88
00:03:14,880 --> 00:03:16,440
no.
Yeah, my gut feeling is no.

89
00:03:16,440 --> 00:03:18,200
I think people will convince
themselves that happens and it

90
00:03:18,200 --> 00:03:20,480
just doesn't.
If you've got a kid, right, it's

91
00:03:20,480 --> 00:03:23,920
just it's that OK, I'm getting
too carried away and calm down a

92
00:03:23,920 --> 00:03:26,320
bit.
If you've got a kid who clearly

93
00:03:26,320 --> 00:03:29,720
can spell age expected for your
six, like their spelling is very

94
00:03:29,720 --> 00:03:32,240
good in the writing, but you
can't find the word soldier in

95
00:03:32,240 --> 00:03:35,440
it or you can't find the word
guarantees not in there.

96
00:03:35,440 --> 00:03:37,240
Like all the government, all
these words on this very

97
00:03:37,240 --> 00:03:38,680
specific list.
Isn't it weird the government

98
00:03:38,680 --> 00:03:40,760
chose government?
Yeah, it's got government in

99
00:03:40,760 --> 00:03:42,120
there.
You can't find hindrance in

100
00:03:42,120 --> 00:03:44,360
there.
It's like you're not going to

101
00:03:44,400 --> 00:03:47,040
go.
Oh yeah, I don't think we, I

102
00:03:47,040 --> 00:03:48,920
don't think they can spell.
You're just going to go, well,

103
00:03:48,920 --> 00:03:52,000
obviously they can spell.
It's just such a such a nonsense

104
00:03:52,000 --> 00:03:52,840
thing.
People worry about it.

105
00:03:52,960 --> 00:03:54,960
And I know the only reason it
gets my back up a little bit is

106
00:03:54,960 --> 00:03:57,160
because I once worked at a
school and it was a long time

107
00:03:57,160 --> 00:04:01,320
ago where they were really,
really, really particular about

108
00:04:01,320 --> 00:04:03,040
that.
And they were saying, we cannot

109
00:04:03,040 --> 00:04:04,680
say they can spell unless
there's like at least three

110
00:04:04,680 --> 00:04:07,160
examples and every bit of
writing of a year 5-6 spelling

111
00:04:07,160 --> 00:04:08,360
word.
So they were, they were

112
00:04:08,360 --> 00:04:11,480
basically shoehorning them in
being like, they need word mats

113
00:04:11,480 --> 00:04:13,840
with them all the time.
You need to, you need to

114
00:04:13,840 --> 00:04:15,840
classify these word mats so that
they're words that would fit

115
00:04:15,840 --> 00:04:17,920
with this genre to really
encourage them to use them.

116
00:04:18,120 --> 00:04:20,160
You need to take them off and
highlight the words as you see

117
00:04:20,160 --> 00:04:21,800
them and then go back.
And then honestly, it got to the

118
00:04:21,800 --> 00:04:23,440
point where they were like, Oh,
you didn't do it for this one.

119
00:04:23,440 --> 00:04:25,240
Can you go back through all of
their writing?

120
00:04:25,560 --> 00:04:27,600
And I'm just write down on this
little list here.

121
00:04:27,600 --> 00:04:29,800
I made an Excel sheet for you,
the words that they did using

122
00:04:29,800 --> 00:04:31,640
that bit of writing and I was
like, what are we doing?

123
00:04:31,640 --> 00:04:35,200
What do you mean?
Firstly, this kid can spell, and

124
00:04:35,200 --> 00:04:37,000
this kid probably won't pass the
spell anything anyway.

125
00:04:37,000 --> 00:04:38,800
Regardless of these words, what
are we talking about?

126
00:04:39,480 --> 00:04:42,280
That is crazy.
And also it's it's just 100

127
00:04:42,280 --> 00:04:43,600
words.
If you've seen the list, you'll

128
00:04:43,600 --> 00:04:46,200
know how random they've seen.
And when we when we make

129
00:04:46,200 --> 00:04:48,600
children shoe wand them into
their writing, we're already

130
00:04:48,600 --> 00:04:50,720
struggling to make their writing
always make sense.

131
00:04:50,880 --> 00:04:53,480
But then you'll have, like, a, a
diary entry of a girl, let's say

132
00:04:53,480 --> 00:04:55,840
a young girl who's had a day at
their aunt's house.

133
00:04:56,080 --> 00:04:57,240
And you'll be reading a
sentence.

134
00:04:57,240 --> 00:04:59,440
And you'll know when a teacher
has said, put in one of these

135
00:04:59,440 --> 00:05:01,280
words because it'll just stick
out like a sore thumb.

136
00:05:01,440 --> 00:05:02,560
They'll say something like,
yeah.

137
00:05:02,560 --> 00:05:04,840
And then my aunt sat me down for
some breakfast, which the

138
00:05:04,840 --> 00:05:06,240
government said has to be
healthy.

139
00:05:06,960 --> 00:05:10,720
Yeah, yeah, they crossed out.
They're like with my with my

140
00:05:10,760 --> 00:05:12,720
three siblings and they've
crossed it down.

141
00:05:12,720 --> 00:05:16,840
But 40, because 40 is one of the
words on my 12th birthday, 12

142
00:05:16,840 --> 00:05:19,200
footer, 12 footer.
My teacher told me it's 12

143
00:05:19,200 --> 00:05:20,880
footer.
On my 12th birthday brackets,

144
00:05:20,880 --> 00:05:22,560
the elf came 12th, the elf came
12th.

145
00:05:22,560 --> 00:05:24,160
That's how I remember to spell
this word, which shows that I

146
00:05:24,160 --> 00:05:27,200
can spell.
It's just nonsense, like we're

147
00:05:27,200 --> 00:05:28,920
losing our minds.
I was shocked to see it.

148
00:05:28,920 --> 00:05:31,240
Basically that's.
Not still OK, so just, I'm just

149
00:05:31,240 --> 00:05:33,080
trying to side with these people
a bit here because there'll be

150
00:05:33,080 --> 00:05:35,040
some really good teachers.
Like, again, earlier in my

151
00:05:35,040 --> 00:05:37,360
career, I'd have fallen for that
trap and been like, oh, OK, you

152
00:05:37,360 --> 00:05:40,160
do need to have these 100
curriculum words that are in

153
00:05:40,160 --> 00:05:42,320
the, in the national curriculum
where.

154
00:05:42,600 --> 00:05:45,400
But I, I want to dig deeper into
it and maybe some people

155
00:05:45,400 --> 00:05:47,760
listening can help us with this
because, you know, do you have

156
00:05:47,760 --> 00:05:50,000
to show evidence of those words
in your children's writing all

157
00:05:50,000 --> 00:05:52,600
the time?
Because if you don't, why are

158
00:05:52,600 --> 00:05:54,800
they even there?
Like surely they're only there

159
00:05:54,800 --> 00:05:57,440
because we do need to evidence
them and show that children can

160
00:05:57,440 --> 00:05:59,160
write them.
Do they come up more often in

161
00:05:59,160 --> 00:06:01,760
the spelling sats?
Do they do do are, are there, is

162
00:06:01,760 --> 00:06:03,560
there a barometer there where
maybe there's more spending

163
00:06:03,560 --> 00:06:05,280
rules involved and therefore
that's why they're good?

164
00:06:05,280 --> 00:06:07,240
Or are they just exception
random words?

165
00:06:07,240 --> 00:06:09,600
What is, what is it?
I think what's happened is

166
00:06:09,600 --> 00:06:11,800
we've, we've just thought about
it the wrong way around.

167
00:06:12,120 --> 00:06:14,920
It's, I think that it's my
understanding that the intention

168
00:06:14,920 --> 00:06:16,640
is, it's not that we have to
shoehorn them in.

169
00:06:16,760 --> 00:06:19,280
It's that when children
naturally write these words, we

170
00:06:19,280 --> 00:06:21,080
should be just checking to see
if that's spelling them right.

171
00:06:21,160 --> 00:06:24,240
So for example, if a, if a kid
happens to have used 5 words

172
00:06:24,240 --> 00:06:26,280
from this list and all five of
them are spelt wrong, then we

173
00:06:26,280 --> 00:06:28,480
can say, right, OK, well,
there's an indicator there that

174
00:06:28,480 --> 00:06:30,520
these words, they've been
exposed to this, this list that

175
00:06:30,520 --> 00:06:32,400
the government have chosen which
kid is happening to write

176
00:06:32,400 --> 00:06:33,960
hindrance.
Yeah, no, I know.

177
00:06:33,960 --> 00:06:35,800
But some of them are like, you
know, some of them are more

178
00:06:35,800 --> 00:06:36,280
shoehorn.
Yeah.

179
00:06:36,280 --> 00:06:38,280
A lot of the words do come up
because they are generally when

180
00:06:38,280 --> 00:06:41,400
you look through the list, they
are generally a lot of them are

181
00:06:41,400 --> 00:06:43,800
just like kind of common.
They're age conception words,

182
00:06:44,160 --> 00:06:46,400
like words where it's like they
do kind of just not follow a

183
00:06:46,400 --> 00:06:49,280
rule, but they're fairly common.
Well, like for example, I know

184
00:06:49,280 --> 00:06:51,600
environments in there and I know
a lot of children do a lot of

185
00:06:51,600 --> 00:06:54,120
work in science and stuff about
the environment and I can see

186
00:06:54,120 --> 00:06:58,120
how that can maybe that's that's
a good example there of how well

187
00:06:58,120 --> 00:07:00,000
that's on the list we're going
to teach about the environment.

188
00:07:00,000 --> 00:07:01,480
Let's just make sure my kids
know how to spell it.

189
00:07:01,680 --> 00:07:04,120
But at the end of the day, if
they finish year 6 and you're

190
00:07:04,120 --> 00:07:06,440
sat in a writing moderation and
you can't find evidence of the

191
00:07:06,440 --> 00:07:08,360
word environment in any.
Of them, you're not going to say

192
00:07:08,360 --> 00:07:10,480
that can't stop.
No, exactly, exactly.

193
00:07:10,480 --> 00:07:12,080
So I think I don't know it's
it's a weird.

194
00:07:12,080 --> 00:07:13,160
I'm sure there's a grey area
there.

195
00:07:13,160 --> 00:07:15,920
Hence like my school did it
years ago to me and but I've

196
00:07:15,920 --> 00:07:17,240
been at so many moderations
since then.

197
00:07:17,240 --> 00:07:19,600
I've grown maybe more confidence
and I've been to internal

198
00:07:19,600 --> 00:07:21,880
moderations in multiple schools
and I've been an external

199
00:07:21,880 --> 00:07:24,520
moderation where you go to like,
you know, work with other

200
00:07:24,520 --> 00:07:27,120
teachers and so on, runs it.
And there's other experts in the

201
00:07:27,120 --> 00:07:28,760
room.
And everyone just unanimously

202
00:07:28,760 --> 00:07:31,080
agrees that you really don't
need to be that picky.

203
00:07:31,400 --> 00:07:34,120
You can just get a general feel
of spellings and just look to

204
00:07:34,120 --> 00:07:35,960
see if they're spelling things
right in general.

205
00:07:35,960 --> 00:07:37,640
And if there's a clearly a
problem with spellings, then

206
00:07:37,680 --> 00:07:38,560
there's a problem with
spellings.

207
00:07:38,560 --> 00:07:40,640
And you can't take off that
objective of like, OK, you know,

208
00:07:40,640 --> 00:07:42,600
there's a lot of year 3 spelling
rules here that are getting

209
00:07:42,600 --> 00:07:45,320
wrong consistently.
Maybe spellings is not something

210
00:07:45,320 --> 00:07:47,400
we can take off for them.
So when you compare it to

211
00:07:47,400 --> 00:07:50,120
something like if the children
aren't consistently using

212
00:07:50,120 --> 00:07:53,240
capital letters for names,
etcetera, that's that can be

213
00:07:53,240 --> 00:07:55,200
something that stops them
getting to age expected.

214
00:07:55,720 --> 00:07:58,840
But you're saying this list
alone should be as like a

215
00:07:58,840 --> 00:08:01,440
supplementary thing that that
doesn't necessarily have the

216
00:08:01,440 --> 00:08:03,560
same weight as OK, they're
missing a word.

217
00:08:03,560 --> 00:08:05,840
It doesn't make sense that can
stop you.

218
00:08:05,920 --> 00:08:09,640
Whereas not spelling soldier
incorrectly once or not

219
00:08:09,640 --> 00:08:11,480
including it at all.
Yeah, that's that.

220
00:08:11,480 --> 00:08:12,960
Shouldn't, yeah, it's not and it
doesn't.

221
00:08:12,960 --> 00:08:15,120
It's not going to happen.
Honestly, in reality, that is

222
00:08:15,120 --> 00:08:17,280
not going to happen if they
spell everything else correctly

223
00:08:17,280 --> 00:08:19,280
and they've got clearly evidence
of let's say we're doing year 6

224
00:08:19,280 --> 00:08:22,240
moderation, clearly got evidence
of tonnes of year 6 spending

225
00:08:22,360 --> 00:08:23,680
rules being spelled fairly
consistent.

226
00:08:23,680 --> 00:08:25,360
It doesn't have to be, it
doesn't have to be every time as

227
00:08:25,360 --> 00:08:27,040
well, just fairly consistently
correctly.

228
00:08:27,240 --> 00:08:29,160
They're going to get through on
the spelling objective and it's

229
00:08:29,160 --> 00:08:31,120
just not, it's just not going to
matter.

230
00:08:31,120 --> 00:08:32,960
I think we just put so much
emphasis on them.

231
00:08:32,960 --> 00:08:35,280
Just to clarify, I'm also siding
with the teachers.

232
00:08:35,280 --> 00:08:37,080
I'm not getting annoyed at the
teachers for asking this.

233
00:08:37,280 --> 00:08:39,200
I'm getting annoyed at the fact
that there are still places

234
00:08:39,200 --> 00:08:42,480
where like they put this on an
insane pedestal.

235
00:08:42,480 --> 00:08:44,360
Yeah, I'm like, why are you
asking me that about?

236
00:08:44,640 --> 00:08:46,960
Your school shouldn't be making
you a half term like feel like

237
00:08:46,960 --> 00:08:49,600
you have to go away and ask for
how to shoehorn in these random

238
00:08:49,600 --> 00:08:51,080
words.
Like if you're going to worry

239
00:08:51,080 --> 00:08:52,800
about anything, it should
probably be of the stuff that

240
00:08:52,800 --> 00:08:55,840
actually matters in terms of
like just how they word, how how

241
00:08:55,840 --> 00:08:58,240
they form their sentences and
grammar and punctuation and

242
00:08:58,240 --> 00:09:00,960
stuff that actually impacts the
flow of writing and how a reader

243
00:09:00,960 --> 00:09:02,480
interprets it.
And inference and

244
00:09:02,480 --> 00:09:04,800
personification, figurative
language, All of the stuff that

245
00:09:04,800 --> 00:09:07,360
like actually matters for being
good at writing.

246
00:09:07,920 --> 00:09:10,600
I tell you what spelling, that
random list of words that the

247
00:09:10,600 --> 00:09:12,520
government have decided is, is
important.

248
00:09:12,520 --> 00:09:16,800
The government, the government
with an N is honestly at the

249
00:09:16,840 --> 00:09:18,960
absolute bottom.
It's the absolute bottom.

250
00:09:18,960 --> 00:09:20,400
It should not be anywhere near
the top.

251
00:09:20,400 --> 00:09:22,440
And it just is for some reason
for some people, so.

252
00:09:22,600 --> 00:09:24,600
Staying on moderation for a
minute, then it's moderation

253
00:09:24,600 --> 00:09:27,560
season, I guess at the moment in
schools and a lot of people be

254
00:09:27,560 --> 00:09:31,920
going through this process.
I always remember deputy head

255
00:09:31,920 --> 00:09:34,680
that I worked with mentioning
how, and you might remember from

256
00:09:34,680 --> 00:09:37,800
school where writing was
assessed for more formally and

257
00:09:37,800 --> 00:09:39,880
then it came in that it was more
teacher assessment and then

258
00:09:39,880 --> 00:09:42,800
therefore moderation came in.
Do you think in general then

259
00:09:42,800 --> 00:09:45,600
based on what you're saying, you
could you could frame what

260
00:09:45,600 --> 00:09:49,440
you're saying to, to argue that
moderation is really important

261
00:09:49,720 --> 00:09:53,120
because we won't have these
people stuck in schools who have

262
00:09:53,120 --> 00:09:55,440
this kind of top down heavy of
this is really important.

263
00:09:55,520 --> 00:09:58,560
When actually it's not like the
only way you might actually find

264
00:09:58,560 --> 00:10:01,520
out that it's not important and
is through moderation, going to

265
00:10:01,520 --> 00:10:03,320
other schools and seeing
experts, etcetera.

266
00:10:03,520 --> 00:10:06,360
So I guess I've got a broad
question to you now as a Year 6

267
00:10:06,360 --> 00:10:10,280
teacher who who teaches writing
to some very capable writers.

268
00:10:10,600 --> 00:10:15,040
Do you think that the moderation
process is is the correct way to

269
00:10:15,040 --> 00:10:18,080
go about assessing children in
their writing?

270
00:10:18,800 --> 00:10:21,800
I think it look, it's a flawed
system.

271
00:10:23,040 --> 00:10:25,920
I don't know if I think I don't
want to say right now, yes, it's

272
00:10:25,920 --> 00:10:28,080
the correct way.
I think it is a fairly good

273
00:10:28,080 --> 00:10:28,760
system.
Is it?

274
00:10:28,760 --> 00:10:31,280
Better than having an assessed
piece that's timed and then just

275
00:10:31,280 --> 00:10:33,960
40 minutes and it goes off to
someone who marks it because I

276
00:10:34,160 --> 00:10:36,440
I'm struggling to see how it
can't be better than that.

277
00:10:37,400 --> 00:10:40,000
I think so, and I'm going to
trust a lot of my elder teachers

278
00:10:40,000 --> 00:10:43,000
who will unanimously also tell
me that moderation is much more

279
00:10:43,000 --> 00:10:45,480
of a fair system to that old
system like that.

280
00:10:45,560 --> 00:10:48,640
So I do think it is, but just
thinking about moderation again,

281
00:10:48,640 --> 00:10:51,240
having had lots of experience
going to moderations, firstly, I

282
00:10:51,240 --> 00:10:55,800
do agree, I think it does allow
a bit more of a like a, rather

283
00:10:55,800 --> 00:10:57,600
than 15 different schools
interpreting something all

284
00:10:57,600 --> 00:10:59,120
separately, right?
It does kind of bring you

285
00:10:59,120 --> 00:11:00,960
together and be like, hang on a
minute, let's just like fresh

286
00:11:00,960 --> 00:11:04,200
page here, let's all agree
together now that this is kind

287
00:11:04,200 --> 00:11:05,120
of the standard we're looking
for.

288
00:11:05,120 --> 00:11:06,840
Now go back to your schools and
make sure like we've got the

289
00:11:06,840 --> 00:11:08,640
same information.
It does kind of level the

290
00:11:08,640 --> 00:11:11,400
playing field a bit in terms of
that, but also people do just

291
00:11:11,400 --> 00:11:13,800
ignore it sometimes.
Not everyone goes to every

292
00:11:13,800 --> 00:11:16,200
moderation.
Sometimes there's an internal

293
00:11:16,200 --> 00:11:17,760
moderation.
So you will just be isolated

294
00:11:17,760 --> 00:11:21,000
with one person coming in.
Is that is that so with an

295
00:11:21,000 --> 00:11:23,560
internal moderation.
So throughout the year, lots of

296
00:11:23,560 --> 00:11:26,120
schools and trusts, etcetera
will set up moderations into

297
00:11:26,120 --> 00:11:28,720
school and you'll kind of go and
there'll be 5 other

298
00:11:28,720 --> 00:11:31,680
representatives represent
representatives from other

299
00:11:31,680 --> 00:11:33,080
schools on the same table as
you.

300
00:11:33,160 --> 00:11:36,040
Maybe you're all in year 4, for
example, you'll bring a book and

301
00:11:36,040 --> 00:11:38,240
you'll say something like, I
think this is expected because

302
00:11:38,240 --> 00:11:39,840
of that.
I don't think this child's quite

303
00:11:39,840 --> 00:11:42,680
expected because of this.
And then everyone kind of brings

304
00:11:42,680 --> 00:11:45,440
what they say is each level and
you look at them and you go,

305
00:11:45,440 --> 00:11:47,000
well, actually that's better
than what I thought.

306
00:11:47,360 --> 00:11:49,480
Oh, no, that all right.
You would you would say this is

307
00:11:49,600 --> 00:11:51,440
expected interesting, because I
wouldn't.

308
00:11:51,760 --> 00:11:53,760
And that is a nice way to
balance the playing field.

309
00:11:53,760 --> 00:11:56,320
I think that's where it's good.
But what I still see a lot

310
00:11:56,320 --> 00:11:59,680
online of, and maybe you could
comment on this, is there's

311
00:11:59,680 --> 00:12:03,640
still huge discrepancies out
there and there's still the the

312
00:12:03,680 --> 00:12:06,800
the thing that gets my back up
about moderation is how easy it

313
00:12:06,800 --> 00:12:09,480
is to fudge things through and
basically cheat because the the

314
00:12:09,480 --> 00:12:11,600
amount of times you've seen
like, you know, moderation where

315
00:12:11,800 --> 00:12:14,440
well, if just if, if this one
gets through, we can argue the

316
00:12:14,440 --> 00:12:15,960
rest.
We'll just make sure if this

317
00:12:15,960 --> 00:12:18,360
person gets through, that gives
us 15 expected and we won't

318
00:12:18,360 --> 00:12:21,720
really look at those or and, you
know, we can talk about cheating

319
00:12:21,720 --> 00:12:24,120
in general aside of what you do
with the children.

320
00:12:24,120 --> 00:12:26,760
But moderation itself can be
kind of a green flag sometimes

321
00:12:27,000 --> 00:12:29,840
to usher your way through and
get your your stats up.

322
00:12:30,000 --> 00:12:31,800
Yeah.
And and also there's people

323
00:12:31,800 --> 00:12:34,480
online who put a piece of
writing up and say, just

324
00:12:34,480 --> 00:12:37,480
wondering, is this greater depth
and it's not even expected?

325
00:12:37,520 --> 00:12:39,240
Yeah, yeah, it's not even at the
expected level.

326
00:12:39,240 --> 00:12:42,160
And so is moderation failing
because we still have people who

327
00:12:42,160 --> 00:12:44,240
clearly, you know, clearly we're
miles off and.

328
00:12:44,680 --> 00:12:46,840
I don't think it's failing.
I think it's, it's a system that

329
00:12:46,840 --> 00:12:49,320
is constantly being fed in new
teachers and old teachers are

330
00:12:49,320 --> 00:12:50,240
leaving.
Do you know what I mean?

331
00:12:50,240 --> 00:12:52,280
Like old, that's horrible
teachers are leaving.

332
00:12:52,280 --> 00:12:54,280
They don't be old, but new
teachers are constantly coming

333
00:12:54,280 --> 00:12:55,160
in.
It might just be they haven't

334
00:12:55,160 --> 00:12:57,080
had experience with it yet.
I think when we see examples of

335
00:12:57,080 --> 00:12:58,840
that, they're probably just less
experienced and this is their

336
00:12:58,840 --> 00:13:00,160
learning curve.
And then maybe they'll go to

337
00:13:00,160 --> 00:13:01,800
moderation at some point.
Maybe it's their first year in

338
00:13:01,800 --> 00:13:04,040
year 6 where they're going to an
official moderation.

339
00:13:04,120 --> 00:13:05,600
I don't know, like I don't think
it's failing.

340
00:13:05,600 --> 00:13:08,560
I think it's probably just
fringe examples where where it,

341
00:13:08,760 --> 00:13:10,800
you know, hasn't caught them yet
Or like you said, it's not

342
00:13:10,800 --> 00:13:12,160
perfect system.
There are still going to be

343
00:13:12,160 --> 00:13:14,880
people that do go to moderation
and ultimately don't get out of

344
00:13:14,880 --> 00:13:15,800
it.
Probably what they should have

345
00:13:15,800 --> 00:13:17,760
got out of it.
But I reckon the vast majority

346
00:13:17,760 --> 00:13:20,400
of people do go to moderation
and come out feeling pretty,

347
00:13:20,760 --> 00:13:22,880
pretty well understood in terms
of what the expectations are.

348
00:13:22,880 --> 00:13:25,960
But going back to one point,
you're so right, because with

349
00:13:26,560 --> 00:13:28,760
with moderation, cross
moderation, we get invited to

350
00:13:28,760 --> 00:13:30,120
it.
It's very official as year six

351
00:13:30,120 --> 00:13:32,640
teachers, we do get invited to
it and it's very much expected

352
00:13:32,640 --> 00:13:34,400
to attend.
You don't have to, but it does

353
00:13:34,400 --> 00:13:36,840
open the door for more scrutiny
and checking with internal

354
00:13:36,840 --> 00:13:39,480
moderations in future years.
It's kind of like come along or

355
00:13:39,480 --> 00:13:41,080
you're going to end up, there's
going to be a consequence.

356
00:13:41,080 --> 00:13:43,160
So you have to pretty much have
to go for people who aren't in

357
00:13:43,160 --> 00:13:45,520
year 6 very quickly.
It's, it's important because

358
00:13:45,520 --> 00:13:48,720
year six teachers will decide
the grade at the end, whereas

359
00:13:48,720 --> 00:13:50,840
other year groups, you can kind
of have a linear approach and

360
00:13:50,880 --> 00:13:52,840
schools can decide.
But for you, it's vital.

361
00:13:52,840 --> 00:13:54,720
Yeah, it's the actual key.
It'll actually be recorded.

362
00:13:54,720 --> 00:13:56,480
The actual end of key stage 2,
right?

363
00:13:56,560 --> 00:13:58,880
Writing assessment is genuinely
down to us and it will go down

364
00:13:58,880 --> 00:14:04,600
on paper, but it just says bring
either five books or 15% of your

365
00:14:04,600 --> 00:14:06,680
cohort if your cohort's over
like a certain number.

366
00:14:06,840 --> 00:14:10,200
So we end up taking like of our
90 odd cohort, we take like 15

367
00:14:10,200 --> 00:14:13,960
odd books and 14/14/15 books.
And it just has to have a, you

368
00:14:13,960 --> 00:14:17,000
know, a variety of levels there.
So most of those books will be

369
00:14:17,000 --> 00:14:18,320
expected.
You'll take a couple of greater

370
00:14:18,320 --> 00:14:19,720
depths because you're
realistically, you're not going

371
00:14:19,720 --> 00:14:21,800
to get loads of them and you'll
take a few you're sort of

372
00:14:21,800 --> 00:14:24,240
working towards or maybe even
below just so you've got a

373
00:14:24,240 --> 00:14:26,160
variety and you can, and you can
go there and then you can say,

374
00:14:26,160 --> 00:14:28,840
right, have we got this right?
And if all 15 books have kind of

375
00:14:28,840 --> 00:14:31,560
been assessed appropriately and,
and people around you agree,

376
00:14:31,680 --> 00:14:33,800
maybe the lead moderator that's
there who kind of wandering

377
00:14:33,800 --> 00:14:35,800
around, they agree of your
gradings as well.

378
00:14:35,800 --> 00:14:37,760
All gets written down.
And generally what happens is if

379
00:14:37,760 --> 00:14:40,640
everyone agrees, you go back,
they're like, OK, we can trust

380
00:14:40,640 --> 00:14:43,560
that school to, to grade the
other 85% of their books.

381
00:14:43,560 --> 00:14:46,080
Yes, crack on, right.
But you've got free choice of

382
00:14:46,080 --> 00:14:47,520
which 15 books.
Yeah, yeah, yeah, yeah.

383
00:14:47,520 --> 00:14:49,720
So, yeah, yeah.
Obviously, and I don't, you

384
00:14:49,720 --> 00:14:52,720
know, you can tell me whether
you what where you think this is

385
00:14:52,720 --> 00:14:55,800
on the scale of cheating to not
Obviously people are naturally

386
00:14:55,800 --> 00:14:57,640
going to pick books they're more
confident in that are actually

387
00:14:57,640 --> 00:14:59,680
going to be upgraded as the
thing they said they are because

388
00:14:59,680 --> 00:15:01,360
you are rewarded for getting the
grading right.

389
00:15:01,440 --> 00:15:02,520
Yeah.
So obviously you're going to

390
00:15:02,680 --> 00:15:03,880
not.
No one's really going there.

391
00:15:03,880 --> 00:15:06,200
The idea is you take your custom
books, you take your custom

392
00:15:06,200 --> 00:15:07,800
books to help work out whether
they are.

393
00:15:07,840 --> 00:15:09,800
Yeah, but really, people just
want to go there.

394
00:15:09,800 --> 00:15:13,800
OK, so to me, right, that is why
I think moderation where it's

395
00:15:13,800 --> 00:15:15,080
pointless.
Yeah, that's the biggest.

396
00:15:15,080 --> 00:15:17,160
So this, this is where I think
there is a huge fall in the

397
00:15:17,160 --> 00:15:20,280
system because it becomes
playing a game and we all sit

398
00:15:20,280 --> 00:15:22,160
there and smile and, and not at
each other as if they say,

399
00:15:22,480 --> 00:15:24,760
aren't we all fantastic at
bringing what is clearly an

400
00:15:24,760 --> 00:15:26,520
expected book to the table?
Isn't that great?

401
00:15:26,800 --> 00:15:29,560
And the government don't help
because the government, what

402
00:15:29,560 --> 00:15:32,120
they give us examples of
expecting and working towards,

403
00:15:32,120 --> 00:15:35,040
etcetera, are very strong,
secure, expected.

404
00:15:35,400 --> 00:15:37,440
So and then, and then the
conversation is, Oh yeah, well,

405
00:15:37,440 --> 00:15:40,080
we don't want to go too close to
the custody line between what is

406
00:15:40,080 --> 00:15:42,600
expected and what isn't because,
you know, we, we might muddy the

407
00:15:42,600 --> 00:15:44,680
waters and there's a, there's a
professional dialogue to have

408
00:15:44,680 --> 00:15:46,840
and you can decide yourself.
I'm like, OK, cool.

409
00:15:46,840 --> 00:15:49,440
Well, if you're not going to
give me that, I'm obviously

410
00:15:49,440 --> 00:15:51,840
going to take my strong books.
This moderation and what we

411
00:15:51,840 --> 00:15:55,880
aren't doing at all, I don't
think efficiently is teaching

412
00:15:55,880 --> 00:16:00,880
staff exactly what makes
something very, very high

413
00:16:00,880 --> 00:16:04,160
working towards nearly there but
not quite and what makes someone

414
00:16:04,200 --> 00:16:04,840
absolutely.
Yeah.

415
00:16:04,840 --> 00:16:06,080
OK.
You just got there, but you are

416
00:16:06,080 --> 00:16:07,040
expecting?
Very low working.

417
00:16:07,120 --> 00:16:11,600
That line exactly that very low
expected that line is not to me

418
00:16:11,840 --> 00:16:15,160
being made very clear to most of
the profession which allows for

419
00:16:15,160 --> 00:16:17,480
this like, well, obviously, OK,
I'm going to bump them up then.

420
00:16:17,640 --> 00:16:19,880
And when I go to moderation, I'm
not going to take the kid I

421
00:16:19,880 --> 00:16:22,720
think who is just about on that
line because what moderation

422
00:16:22,720 --> 00:16:24,800
should be is you take your
lowest child of what you think

423
00:16:24,800 --> 00:16:27,560
has achieved expected.
It gets nodded through maybe

424
00:16:27,560 --> 00:16:28,960
like, OK, yeah, do you know what
just about.

425
00:16:29,200 --> 00:16:32,240
And then that does give you the
green flag to everyone above

426
00:16:32,240 --> 00:16:34,200
that to be like, well, if this
is then that's great.

427
00:16:34,320 --> 00:16:36,640
What's actually happening is
you're taking your most secure

428
00:16:36,640 --> 00:16:39,880
expected children and then going
down the line and being like,

429
00:16:39,880 --> 00:16:42,640
well, because that because this
person who never no one would

430
00:16:42,640 --> 00:16:45,200
ever disagree is expected has
been agreed to be expected.

431
00:16:45,200 --> 00:16:46,240
Who would ever have thought
that?

432
00:16:46,240 --> 00:16:48,200
Wow.
Because of that, everyone I've

433
00:16:48,200 --> 00:16:53,200
put expected now is that is not,
to me, an accurate use or a good

434
00:16:53,200 --> 00:16:55,880
use of moderation time because
we're jumping through hoops.

435
00:16:55,880 --> 00:16:58,600
We're trying to get data, we're
trying to boost our scores more

436
00:16:58,600 --> 00:17:01,560
than we are actually trying to
properly decide where children

437
00:17:01,560 --> 00:17:02,320
are on the.
Scale.

438
00:17:02,520 --> 00:17:04,560
And that's why I think it's a
flawed system and I don't want

439
00:17:04,560 --> 00:17:06,640
to discredit everyone because
I'm sure some people do actually

440
00:17:06,640 --> 00:17:09,839
go with the proper intent of I
am actually going to take the

441
00:17:09,839 --> 00:17:11,280
kids that I'm not too sure of
and I'm going to use the

442
00:17:11,280 --> 00:17:12,599
moderation process to get this
right.

443
00:17:12,599 --> 00:17:15,440
But then, but then those people
get punished because other

444
00:17:15,440 --> 00:17:17,800
people on the table will also
say, OK, these are all my

445
00:17:17,800 --> 00:17:20,720
expected children and that
someone else has brought someone

446
00:17:20,720 --> 00:17:24,359
who's solid expected, who
clearly is gets compared then to

447
00:17:24,359 --> 00:17:26,720
your one who's Cusby.
And what's more likely to happen

448
00:17:26,720 --> 00:17:28,720
is say, well, they're not there.
So probably that's working

449
00:17:28,720 --> 00:17:31,520
towards because unless
everyone's doing it properly,

450
00:17:31,720 --> 00:17:34,320
you're gonna be punished for
bringing your lower children in

451
00:17:34,320 --> 00:17:35,160
that moderation.
Process.

452
00:17:35,160 --> 00:17:37,840
Yeah, I agree.
And why ultimately does this

453
00:17:37,840 --> 00:17:41,760
happen because there is pressure
to hit data standards from the

454
00:17:41,760 --> 00:17:43,080
government.
The government are saying well

455
00:17:43,080 --> 00:17:45,360
national average is 72% is
inflated every year by the

456
00:17:45,360 --> 00:17:47,840
previous year, sure.
Then our national standards are

457
00:17:47,840 --> 00:17:52,120
74% of children are are expected
in writing so you better hit

458
00:17:52,120 --> 00:17:53,760
that.
So OK, well, if you're telling

459
00:17:53,760 --> 00:17:55,520
that to teachers, oh, by the
way, your pay is performance

460
00:17:55,520 --> 00:17:57,240
related.
OK, well it's funny that

461
00:17:57,240 --> 00:17:59,600
everyone's suddenly getting 74
plus percent, isn't it?

462
00:17:59,720 --> 00:18:02,160
You know, I'm obviously knocks,
that's how averages work, but.

463
00:18:02,240 --> 00:18:04,040
Yeah, yeah.
I don't know, I think that the

464
00:18:04,200 --> 00:18:05,520
the top down pressure doesn't
help.

465
00:18:06,000 --> 00:18:09,360
I think, I think moderation
itself could be more straight in

466
00:18:09,360 --> 00:18:10,880
terms of the rules of like how
it works.

467
00:18:10,920 --> 00:18:12,720
It might be able to do
moderation, yeah, rather than

468
00:18:12,720 --> 00:18:15,360
just saying bring books and oh,
you know.

469
00:18:15,560 --> 00:18:17,360
Try on the back.
Try on the back, try and bring

470
00:18:17,360 --> 00:18:18,560
books that around, you know, the
cusp.

471
00:18:18,840 --> 00:18:20,120
Oh yeah, cool.
Everyone's like, yeah, OK, cool,

472
00:18:20,120 --> 00:18:21,280
I'll bring books around the cusp
level.

473
00:18:21,520 --> 00:18:23,560
Bloody not gonna make sure my
date is good because I'm gonna

474
00:18:23,800 --> 00:18:26,120
I'm in an Academy trust and
they're not gonna pay me next

475
00:18:26,120 --> 00:18:27,560
year unless I get this hit this
target.

476
00:18:28,160 --> 00:18:31,280
Yeah, what it could, you know,
maybe, maybe moderation could be

477
00:18:31,280 --> 00:18:33,840
redesigned a little bit where
it's like, no, no, no, this game

478
00:18:33,840 --> 00:18:36,320
is like you're bringing way
fewer books now everyone has to

479
00:18:36,320 --> 00:18:39,360
bring you have to bring your top
working towards and your lowest

480
00:18:39,360 --> 00:18:40,240
expected.
You have to.

481
00:18:40,600 --> 00:18:42,720
And if you get it vastly wrong,
if you're bringing your best

482
00:18:42,720 --> 00:18:44,880
expected and trying to pretend
that it's lowest expected, it'll

483
00:18:44,880 --> 00:18:47,120
be obviously get called out.
It's like, OK, well you're

484
00:18:47,120 --> 00:18:48,640
great.
It always backfires.

485
00:18:48,640 --> 00:18:50,600
Like if you think that's your
lowest expected, you're not,

486
00:18:50,680 --> 00:18:52,760
you're not very secure of the
teacher assessment framework,

487
00:18:52,960 --> 00:18:54,720
which means your gradings are
probably quite wrong.

488
00:18:55,200 --> 00:18:56,840
So I guess we need to do
internal moderation with you

489
00:18:56,840 --> 00:18:59,840
because even if that's wrong, so
it's like being held more

490
00:18:59,840 --> 00:19:02,720
accountable maybe to like the
very strict parameters could be

491
00:19:02,720 --> 00:19:04,680
a good way of identifying
whether teachers, because the

492
00:19:04,680 --> 00:19:06,160
whole point is, do they
understand the teacher

493
00:19:06,160 --> 00:19:08,440
assessment framework and are
therefore assessing their

494
00:19:08,440 --> 00:19:10,720
children's books correctly?
That's all we want to know at

495
00:19:10,720 --> 00:19:12,840
moderation.
From a moderation's perspective,

496
00:19:12,840 --> 00:19:15,360
are schools doing it properly?
Because you can't go, you can't

497
00:19:15,360 --> 00:19:17,360
go in.
When we say internally moderate,

498
00:19:17,360 --> 00:19:20,800
we mean like an external source
coming into a school, moderating

499
00:19:20,800 --> 00:19:23,000
that school in particular.
And that's a really stressful

500
00:19:23,000 --> 00:19:24,920
time.
Actually, if you get chosen for

501
00:19:24,920 --> 00:19:27,960
that internal moderation, it is
stressful because then you are

502
00:19:27,960 --> 00:19:31,360
under much more of a microscope.
So you've got to ask yourself

503
00:19:31,360 --> 00:19:33,920
the question, why is that more
stressful than the other

504
00:19:33,920 --> 00:19:35,880
moderation?
It's because it's more of a

505
00:19:35,880 --> 00:19:37,400
moderation.
Yeah, because the other

506
00:19:37,400 --> 00:19:39,640
moderation, which is just
between schools and the choice

507
00:19:39,880 --> 00:19:42,640
is much more laid back, is much
more backpacking, is much more

508
00:19:42,640 --> 00:19:44,120
saying, OK, cool, aren't we all
great?

509
00:19:44,520 --> 00:19:47,000
That internal moderation, the
reason people get their cackles

510
00:19:47,000 --> 00:19:49,400
up if they get chosen, is
because they know they're about

511
00:19:49,400 --> 00:19:51,440
to be exposed as inflating the
scores.

512
00:19:51,440 --> 00:19:52,680
Yeah.
Also, you can't pick your books,

513
00:19:52,880 --> 00:19:54,080
right?
When it's when internal

514
00:19:54,080 --> 00:19:55,840
moderators come through, you
give them a class list and they

515
00:19:55,840 --> 00:19:57,720
just pick randomly kids.
So if they pick a kid that you

516
00:19:57,720 --> 00:19:59,280
wouldn't have taken the
moderation because you're like,

517
00:19:59,280 --> 00:20:00,960
oh, that's one of my custody
ones that I want to put as

518
00:20:00,960 --> 00:20:02,200
expected, but someone might
argue.

519
00:20:02,280 --> 00:20:04,800
It down and they pick it then
tough luck mate, because you've

520
00:20:04,800 --> 00:20:07,120
already put it as expected.
And if they knock it down, yeah,

521
00:20:07,240 --> 00:20:08,400
if they knock it down, they have
to.

522
00:20:08,400 --> 00:20:11,600
It triggers like if they do it
with a few books, yeah, it

523
00:20:11,600 --> 00:20:13,480
triggers them having to look at
every single book, yeah.

524
00:20:13,640 --> 00:20:16,680
Yeah, so and what's gonna happen
when they do pick a random book

525
00:20:16,680 --> 00:20:18,360
and it happens to be a random
book you've forgotten to take

526
00:20:18,360 --> 00:20:20,160
your post it notes out of where
you've completely cheated and

527
00:20:20,160 --> 00:20:22,480
told the children what to write.
What happens then?

528
00:20:23,000 --> 00:20:24,320
Oh.
Yeah, just very quickly last

529
00:20:24,320 --> 00:20:25,920
thing on this point about
writing and stuff because it

530
00:20:25,920 --> 00:20:28,800
does really annoy me the fact
we've all kind of accepted and

531
00:20:28,800 --> 00:20:32,560
it's just normal to say, yeah,
just do feedback with post it

532
00:20:32,560 --> 00:20:34,520
notes so we can take it out and
no one will see it at the end

533
00:20:35,080 --> 00:20:36,120
that is.
Cheating.

534
00:20:36,120 --> 00:20:37,480
Yeah, it's crazy how it's
cheating.

535
00:20:37,840 --> 00:20:39,680
Some people might be like, don't
do that in my school.

536
00:20:39,680 --> 00:20:43,360
Okay, maybe you don't, but I
tell you what, we speak to a lot

537
00:20:43,360 --> 00:20:46,720
of teachers now and we get
stories all the time about stuff

538
00:20:46,720 --> 00:20:47,840
like this.
This is happening in this

539
00:20:47,840 --> 00:20:49,040
school.
I worked in a school that did

540
00:20:49,040 --> 00:20:50,280
this.
I know someone who works in a

541
00:20:50,280 --> 00:20:53,440
school, did this all the time,
about how they just cheat, flirt

542
00:20:53,440 --> 00:20:54,440
around.
No, no, no, no.

543
00:20:54,440 --> 00:20:54,960
Cheat.
No.

544
00:20:54,960 --> 00:20:56,520
Yeah, yeah, yeah.
This is what this is.

545
00:20:56,800 --> 00:20:58,880
I'm, I'm not, I'm not annoyed at
you or anything.

546
00:20:58,880 --> 00:21:01,400
I'm just saying in general, we
do consistently cheat as a

547
00:21:01,440 --> 00:21:03,280
profession.
Yeah, we we absolutely do.

548
00:21:03,280 --> 00:21:05,680
Because as soon as you're
saying, oh, it's OK, put it on

549
00:21:05,680 --> 00:21:07,920
the post it note.
So it's not in the book, you are

550
00:21:07,920 --> 00:21:10,360
by definition saying it's
because you know that's cheating

551
00:21:10,360 --> 00:21:12,280
to do that.
And you know by definition that

552
00:21:12,440 --> 00:21:15,560
if that exact same thing you've
done is witnessed by someone, it

553
00:21:15,560 --> 00:21:17,800
would then be not an independent
piece of writing.

554
00:21:18,000 --> 00:21:21,480
So you when, if, if you know of
anyone or a school who is doing

555
00:21:21,480 --> 00:21:23,640
that kind of thing where there
are post it notes going into

556
00:21:23,640 --> 00:21:26,520
work to rewrite stuff or
teachers sat with children to

557
00:21:26,520 --> 00:21:27,880
say, no, no, rewrite this, this
is wrong.

558
00:21:27,880 --> 00:21:30,520
Because of that, it's not
independent in writing anymore.

559
00:21:30,720 --> 00:21:33,360
It can't be used in moderation
and it doesn't show that child's

560
00:21:33,360 --> 00:21:35,640
ability in writing.
Whether we want to say it's

561
00:21:35,640 --> 00:21:38,520
because of external pressures or
not or things from the top, I'm

562
00:21:38,520 --> 00:21:40,880
all I'm all ears and we should
maybe change that.

563
00:21:41,320 --> 00:21:44,120
But it But please call a spade a
spade, because that is.

564
00:21:44,120 --> 00:21:46,520
Cheating, Yeah, fair play.
Can we finish this conversation

565
00:21:46,520 --> 00:21:48,800
with one little Horror Story?
Just just my own experience is

566
00:21:48,800 --> 00:21:50,840
not to do not to do with that.
I'm not going to sit here and be

567
00:21:50,880 --> 00:21:53,120
like I was cheating and the
thing I'm not going to go.

568
00:21:53,120 --> 00:21:55,640
That's just not what's happened.
But there are just a Horror

569
00:21:55,640 --> 00:21:56,920
Story of moderation in general,
right?

570
00:21:57,600 --> 00:21:59,920
With a school I worked in again
many, many years ago, right.

571
00:22:00,240 --> 00:22:02,960
And to be given credit, it was
all fairly new for them as well.

572
00:22:02,960 --> 00:22:05,000
It was a fairly new process.
It was a new cohort or whatever

573
00:22:05,000 --> 00:22:07,000
new year says, quite a new
curriculum, I suppose, quite a

574
00:22:07,000 --> 00:22:08,520
new curriculum.
It was a fairly new school in

575
00:22:08,520 --> 00:22:10,360
all this case or whatever, and
all people were nice and

576
00:22:10,360 --> 00:22:12,480
well-intentioned.
But it did end up getting to the

577
00:22:12,480 --> 00:22:14,920
point where we had an internal
moderation booked in.

578
00:22:15,560 --> 00:22:17,320
We knew someone was coming in.
We knew it was the scary

579
00:22:17,320 --> 00:22:19,600
version.
I was the E6 teacher and I

580
00:22:19,600 --> 00:22:24,360
basically just got coerced into
taking all the books home and

581
00:22:24,360 --> 00:22:27,200
having to sit there with an
open, open document and have to

582
00:22:27,200 --> 00:22:31,960
make honestly, like a page table
per piece of writing per kid for

583
00:22:31,960 --> 00:22:34,960
the last like 5 bits of writing
to show all of the evidence of

584
00:22:34,960 --> 00:22:36,200
all the stuff that they could
do.

585
00:22:36,600 --> 00:22:37,880
Right.
And I remember saying to them,

586
00:22:38,160 --> 00:22:40,000
this seems really pointless
because the moderator will just

587
00:22:40,000 --> 00:22:41,120
look for the book and find it
themselves.

588
00:22:41,120 --> 00:22:42,880
That's like, that's like the
entire point of their job.

589
00:22:43,040 --> 00:22:45,280
Why are we providing them like
this list of like exactly where

590
00:22:45,280 --> 00:22:46,400
to find it?
And they were, they were like,

591
00:22:46,400 --> 00:22:48,280
oh, it's, you know, firstly,
it's a good way for us to check

592
00:22:48,280 --> 00:22:49,600
that these books are definitely
expected.

593
00:22:49,600 --> 00:22:51,080
I was like, OK, fair enough.
You're making sure our

594
00:22:51,320 --> 00:22:53,240
assessments are right by going
through the bits of writing,

595
00:22:53,240 --> 00:22:55,200
making sure we definitely have
ticked off all these things.

596
00:22:56,520 --> 00:22:59,600
But come on.
And I spent my whole half term,

597
00:22:59,600 --> 00:23:03,520
not joking, like 2-3 hours per
book per kid where I was going

598
00:23:03,520 --> 00:23:06,360
through that blast, like 5 bits
of writing, having to write down

599
00:23:06,360 --> 00:23:08,400
this table.
And one of the columns was EFI 6

600
00:23:08,400 --> 00:23:09,640
spelling words, what words
they've used.

601
00:23:09,640 --> 00:23:11,920
One of them was I just had to
take off every single objective,

602
00:23:12,120 --> 00:23:13,840
whether that objective was
ticked off in that bit of

603
00:23:13,840 --> 00:23:16,760
writing.
And it took me days and days and

604
00:23:16,760 --> 00:23:18,240
days of my term.
And I really didn't want to do

605
00:23:18,240 --> 00:23:19,120
it.
But they were at the point where

606
00:23:19,120 --> 00:23:21,520
they were basically like, I felt
like I was doing a terrible job

607
00:23:21,520 --> 00:23:23,200
if I didn't do this.
Like I was an awful teacher

608
00:23:23,200 --> 00:23:25,440
because they were putting so
much pressure on and then do you

609
00:23:25,440 --> 00:23:26,480
know what happened in
moderation?

610
00:23:26,560 --> 00:23:30,280
Exactly what I told them would
happen in moderation 5000 times

611
00:23:30,280 --> 00:23:32,560
before that, which was the
moderator won't even pick these

612
00:23:32,560 --> 00:23:34,600
up.
Didn't sat there in a pile the

613
00:23:34,600 --> 00:23:36,520
whole time.
My a whole half term just sat

614
00:23:36,520 --> 00:23:38,800
there in this pile.
Even said oh, we've got this and

615
00:23:38,800 --> 00:23:40,600
said oh, OK, yeah, let's have a
look at the book.

616
00:23:40,680 --> 00:23:41,880
Spent 3 minutes looking at the
book.

617
00:23:41,880 --> 00:23:43,360
Yeah, obviously expected.
Yeah.

618
00:23:43,560 --> 00:23:44,320
Exactly.
Cool.

619
00:23:44,320 --> 00:23:46,160
OK, so you can just get a feel.
Yeah, of course I can get a

620
00:23:46,160 --> 00:23:47,240
feel.
I'm a moderator and I've looked

621
00:23:47,240 --> 00:23:49,480
at thousands of books.
I I can see when a kid is

622
00:23:49,560 --> 00:23:52,680
clearly expected because do you
know what the the range of being

623
00:23:52,680 --> 00:23:55,720
expected is freaking huge.
So when they're up in this top

624
00:23:55,720 --> 00:23:57,960
range, it's really obvious
immediately that they're there.

625
00:23:57,960 --> 00:23:59,880
You don't even have to sit there
and go from every single

626
00:23:59,880 --> 00:24:01,200
objective.
It's so obvious.

627
00:24:01,400 --> 00:24:03,720
So as a trained moderator, you
go, OK, that's clearly expected.

628
00:24:03,720 --> 00:24:05,040
Well, maybe should we check for
greater depth?

629
00:24:05,160 --> 00:24:06,360
Let's have a look.
No, they haven't really hit

630
00:24:06,360 --> 00:24:08,160
these standards of greater
depth, but very strong expected.

631
00:24:08,160 --> 00:24:10,000
Well done.
There's my 3 hours of work on

632
00:24:10,000 --> 00:24:11,920
that child going through and
writing out every single time

633
00:24:11,920 --> 00:24:13,440
they wrote the word government
and hindrance in there.

634
00:24:14,520 --> 00:24:16,720
Oh my God, like that to me, I
quit that school.

635
00:24:17,320 --> 00:24:19,640
I'd already quit by that point
because I could see how I was

636
00:24:19,640 --> 00:24:21,680
just going to hate working in
conditions like that because it

637
00:24:21,680 --> 00:24:24,280
was just absence of common sense
and I just couldn't do it.

638
00:24:24,280 --> 00:24:27,520
I just couldn't do it.
That is the killer, isn't it?

639
00:24:27,520 --> 00:24:29,160
That is the killer in school, I
must say.

640
00:24:30,200 --> 00:24:32,480
You know, I've, I've not got the
most experience in the world

641
00:24:32,480 --> 00:24:34,520
about other workplaces and
stuff, but I have worked in

642
00:24:34,520 --> 00:24:36,080
other workplaces beforehand for
a bit.

643
00:24:36,320 --> 00:24:40,840
I did AI did a couple of weeks.
I got, you know, headhunted

644
00:24:40,840 --> 00:24:43,440
classic like out of graduate
scheme to come and do this thing

645
00:24:43,440 --> 00:24:48,160
and, and the job was basically
head hunting for, for tech firms

646
00:24:48,160 --> 00:24:50,280
and getting people in.
I, I did all that for a few

647
00:24:50,280 --> 00:24:51,920
weeks up in London.
I travelled in and stuff.

648
00:24:52,160 --> 00:24:55,200
And I, I, I just think from what
I've seen and what I've heard

649
00:24:55,200 --> 00:25:01,120
from people, the, the amount of
in education, the stuff we do

650
00:25:01,120 --> 00:25:05,120
that is so ridiculously
pointless, like what you've just

651
00:25:05,120 --> 00:25:09,120
said, utterly, completely
absolute waste of time.

652
00:25:10,000 --> 00:25:12,440
The amount of stuff that we do
in education, which falls into

653
00:25:12,440 --> 00:25:14,880
that bracket is completely
insane.

654
00:25:15,040 --> 00:25:17,600
And This is why I'm really
passionate about when we do this

655
00:25:17,600 --> 00:25:20,920
podcast and when we talk about
this stuff, because the idea of

656
00:25:20,920 --> 00:25:24,520
having Someone Like You, what
you just said in the air of

657
00:25:24,520 --> 00:25:27,240
someone now who's 2-3 years in
the same position, being forced

658
00:25:27,240 --> 00:25:29,200
to do something.
And you're there saying, guys,

659
00:25:29,200 --> 00:25:31,120
I've come out the other side.
And you are right.

660
00:25:31,120 --> 00:25:32,680
It is pointless.
Yeah.

661
00:25:32,800 --> 00:25:35,680
It to know that sometimes just
because someone's in a position

662
00:25:35,680 --> 00:25:38,800
of power over you in work does
not automatically mean that they

663
00:25:38,800 --> 00:25:41,960
know what they're talking about.
Because in that situation, you

664
00:25:41,960 --> 00:25:43,120
were right.
And it didn't matter.

665
00:25:43,120 --> 00:25:45,280
You were three years into your
career and they were 74 years

666
00:25:45,280 --> 00:25:47,640
into their career.
It didn't matter that they could

667
00:25:47,640 --> 00:25:49,160
try and justify it in some way.
Yeah.

668
00:25:49,200 --> 00:25:50,880
Yeah.
The argument was so balanced to

669
00:25:50,880 --> 00:25:52,960
your side, but you still did it
because you felt pressured into

670
00:25:52,960 --> 00:25:54,280
doing it.
Guys, don't feel pressured into

671
00:25:54,280 --> 00:25:56,600
doing anything like.
That I wish I had the Gump back

672
00:25:56,600 --> 00:25:58,520
then to, to have this voice I
have now.

673
00:25:58,520 --> 00:25:59,960
Because I knew what I what I
just said there.

674
00:25:59,960 --> 00:26:01,920
I knew that then.
I knew it, but I just wasn't as

675
00:26:01,920 --> 00:26:03,160
confident.
So what's the only difference

676
00:26:03,160 --> 00:26:05,280
between then and now for you?
Is it that confidence to be like

677
00:26:05,280 --> 00:26:08,080
I know my right to know I've is
it experience?

678
00:26:08,080 --> 00:26:09,080
Does it just come with
experience?

679
00:26:09,080 --> 00:26:10,680
For me it's.
Experience and because I because

680
00:26:10,680 --> 00:26:13,240
I have been through, I guess now
I can look back and be like, I

681
00:26:13,240 --> 00:26:15,200
literally know it's not even a
guess anymore.

682
00:26:15,200 --> 00:26:17,080
I know this is pointless.
And I tell you what, if you

683
00:26:17,080 --> 00:26:18,960
don't believe me, let's go back
in time and I'll go show them

684
00:26:18,960 --> 00:26:20,640
what I did and be like, look,
none of these got looked at.

685
00:26:20,840 --> 00:26:23,640
Just so you know, but and also,
like you said, just like, you

686
00:26:23,640 --> 00:26:27,000
know, I like to think that if,
if this, what we've produced

687
00:26:27,000 --> 00:26:30,040
here existed and I was telling
myself this, that that that

688
00:26:30,040 --> 00:26:32,320
amount of time ago, I might
genuinely have not done it if I

689
00:26:32,320 --> 00:26:33,880
had to have had a bit back up.
I was alone.

690
00:26:33,880 --> 00:26:36,800
I was completely alone.
No one was on my side.

691
00:26:36,800 --> 00:26:38,520
I didn't have anyone backing up
my voice.

692
00:26:39,240 --> 00:26:40,840
I was just alone, so I fell for
the pressure.

693
00:26:40,880 --> 00:26:42,480
Well, guys, look, you're not
alone.

694
00:26:42,480 --> 00:26:45,000
You've got us, Dylan and Hayden
to talk you through all this

695
00:26:45,000 --> 00:26:46,800
stuff.
We're going to go now and talk

696
00:26:46,800 --> 00:26:48,520
about after that lovely chat
about moderation.

697
00:26:48,720 --> 00:26:52,480
I think if I were to define 2,
two things that that are really

698
00:26:52,480 --> 00:26:55,040
important at this part of the
year and take up so much time

699
00:26:55,040 --> 00:26:57,160
for teachers, it is writing
moderation that we've just

700
00:26:57,160 --> 00:26:59,200
spoken about and it is writing
reports.

701
00:26:59,400 --> 00:27:01,720
So let's dive into it, Look a
little bit AI in general and how

702
00:27:01,720 --> 00:27:03,920
it can help and just that kind
of whole discussion.

703
00:27:03,920 --> 00:27:08,440
I'm really excited if you've got
this fine to the episode which

704
00:27:08,440 --> 00:27:10,040
you have because you're
listening to me right now and

705
00:27:10,040 --> 00:27:11,920
you haven't left us a review
yet.

706
00:27:11,920 --> 00:27:14,400
Wherever you're listening to us,
where it's Spotify, Apple

707
00:27:14,400 --> 00:27:17,000
podcast, all that good stuff.
It really helps us out if you

708
00:27:17,000 --> 00:27:19,720
can just leave us a five star
review right now.

709
00:27:19,720 --> 00:27:22,520
Because also I think a big thing
that we're going to notice in

710
00:27:22,520 --> 00:27:25,040
the next few years, all podcasts
will be completely dead because

711
00:27:25,040 --> 00:27:27,320
AI is going to take over and
there won't be here we go

712
00:27:27,520 --> 00:27:29,840
anyway, here we go.
So whilst we can, if you can

713
00:27:29,840 --> 00:27:32,280
give us a review guys, we would
really appreciate it.

714
00:27:32,280 --> 00:27:34,760
Please, let's get to 300 before
the bots take over.

715
00:27:35,680 --> 00:27:38,240
Wow.
Well, that's that's at the tone,

716
00:27:38,240 --> 00:27:39,360
didn't it?
That's my segue.

717
00:27:39,480 --> 00:27:41,040
Yeah, yeah.
Should we talk about AI Dylan?

718
00:27:41,080 --> 00:27:43,720
Yeah, I think let's just root it
in the fact that we're going to

719
00:27:43,720 --> 00:27:46,760
be talking about reports and a
big thing in the last few years.

720
00:27:46,760 --> 00:27:50,120
We've done an episode I think
every year since 2023.

721
00:27:50,160 --> 00:27:52,640
We did one in 2024 and this time
we're doing one now.

722
00:27:52,840 --> 00:27:56,640
Because I do think there is a
huge force for good in using AI

723
00:27:57,000 --> 00:28:01,080
to help save time, in particular
parts of teaching and especially

724
00:28:01,200 --> 00:28:05,160
in report writing when we are
making big huge reams of text.

725
00:28:05,160 --> 00:28:07,360
We're going to, we get into that
and why we think it's a good

726
00:28:07,360 --> 00:28:09,440
thing, the drawbacks, etcetera.
And have a bit of a debate.

727
00:28:10,040 --> 00:28:13,520
But I think first I want to talk
about AI in general in

728
00:28:13,520 --> 00:28:16,240
education.
And then even before that, I

729
00:28:16,240 --> 00:28:19,080
want to take a meta look at the
minute very quickly about AI.

730
00:28:19,080 --> 00:28:20,520
Because there'll be lots of
people listening.

731
00:28:20,680 --> 00:28:22,560
There'll be some people who are
really down a rabbit hole with

732
00:28:22,560 --> 00:28:25,120
what AI is doing at the minute.
And there'll be some people who

733
00:28:25,120 --> 00:28:28,360
still just see AI as ChatGPT
writing me a waggle, like what a

734
00:28:28,360 --> 00:28:30,760
good one looks like in my class
with some determiners in there.

735
00:28:31,400 --> 00:28:34,200
Hayden, can you just try and
summarise briefly for people

736
00:28:34,200 --> 00:28:39,000
listening right now where we're
at in terms of AI and, and what

737
00:28:39,000 --> 00:28:42,520
it's doing and why?
Actually, there's a little bit,

738
00:28:42,760 --> 00:28:44,600
we feel a bit of trepidation
about it.

739
00:28:44,600 --> 00:28:47,840
Yeah, OK, So large language
models, what everyone thinks of

740
00:28:47,880 --> 00:28:50,480
ChatGPT in its infancy.
When we first got it a couple of

741
00:28:50,480 --> 00:28:53,040
years ago, it was just a really
good large language model.

742
00:28:53,240 --> 00:28:55,520
It was very good at predicting
how to say things to us.

743
00:28:55,520 --> 00:28:57,080
And everyone was like, oh, it
seems so human.

744
00:28:57,200 --> 00:28:58,480
We were like, no, it's not
human.

745
00:28:58,480 --> 00:29:01,200
It just knows how to predict the
best words in order to say to

746
00:29:01,200 --> 00:29:02,080
us.
That's all it is.

747
00:29:02,640 --> 00:29:05,520
Couple of years down the line, a
lot of money's been pumped into

748
00:29:05,600 --> 00:29:07,640
AI.
It's getting insanely good at

749
00:29:07,640 --> 00:29:09,720
reasoning now.
It's getting considerably better

750
00:29:09,720 --> 00:29:11,520
at solving problems and
logistics and reasoning.

751
00:29:11,680 --> 00:29:13,760
It's in the background.
It's gone down this whole lines

752
00:29:13,760 --> 00:29:15,240
of like developing code and
developing apps.

753
00:29:15,240 --> 00:29:17,480
And we're at the point now where
it's even getting to the idea of

754
00:29:17,480 --> 00:29:19,680
like, oh, we can actually
replace entire jobs with AI now.

755
00:29:19,880 --> 00:29:22,040
We can make things called AI
agents, which is essentially a

756
00:29:22,040 --> 00:29:24,440
huge workflow of lots of AI
doing loads of different things

757
00:29:24,640 --> 00:29:28,000
into one culmination of, oh,
this is basically what a job is.

758
00:29:28,000 --> 00:29:31,720
And importantly, the thing with
an AI agent is where it used to

759
00:29:31,720 --> 00:29:33,920
be a language model, where it
predict text or use the Internet

760
00:29:33,920 --> 00:29:37,000
to find information, etc, but it
would just do that job.

761
00:29:37,040 --> 00:29:40,000
Yeah, with an AI model, I'm
right in thinking that there's

762
00:29:40,000 --> 00:29:42,160
reasoning involved.
There's thinking involved.

763
00:29:42,160 --> 00:29:43,760
There's a lot of reasoning
involved and there's a lot of

764
00:29:44,120 --> 00:29:45,840
collaboration with other
applications.

765
00:29:45,840 --> 00:29:47,360
Sure, that's what it can do
really, really well now.

766
00:29:47,360 --> 00:29:51,200
So it's the point where people
can build an AI agent to do

767
00:29:51,360 --> 00:29:53,000
honestly, to do an entire like a
marketing job.

768
00:29:53,160 --> 00:29:55,920
You can click a button and it
will go for this huge workflow

769
00:29:55,920 --> 00:29:57,280
of things.
And it will and it will write a

770
00:29:57,280 --> 00:29:58,640
Gmail and it will send the
Gmail.

771
00:29:58,640 --> 00:30:01,440
It will collect the data.
It will then create because it's

772
00:30:01,440 --> 00:30:02,760
because image generation is
really good now.

773
00:30:02,760 --> 00:30:04,080
It will then create an image
post.

774
00:30:04,080 --> 00:30:05,720
It will post it on your
platforms because you've given

775
00:30:05,720 --> 00:30:07,680
it access to do that.
It will harvest the data.

776
00:30:07,680 --> 00:30:09,960
It will repost good ones.
Like you can build a whole

777
00:30:09,960 --> 00:30:12,280
machine, which is someone's job
out of AI now.

778
00:30:12,560 --> 00:30:14,960
And it's crazy, the.
Stuff isn't there like where you

779
00:30:14,960 --> 00:30:17,280
know you can have it like, for
example, weather dependent and

780
00:30:17,280 --> 00:30:20,720
like if if raining in this area,
put a post out about selling my

781
00:30:20,720 --> 00:30:22,960
umbrellas and things like let's
just say for example.

782
00:30:23,080 --> 00:30:25,120
So that's where the reasoning
comes in like it will it will

783
00:30:25,120 --> 00:30:27,440
firstly check the weather if
it's not raining, it won't put

784
00:30:27,440 --> 00:30:29,800
any money out into ads That
where this is a very crude

785
00:30:29,800 --> 00:30:31,960
example but where if it's
raining someone might see an

786
00:30:31,960 --> 00:30:33,480
advert and go oh, I do need this
thing yeah.

787
00:30:34,000 --> 00:30:36,640
It it it does all that reasoning
for you and it's built in and

788
00:30:36,640 --> 00:30:39,520
it's completely insane.
And something I saw recently

789
00:30:39,720 --> 00:30:42,360
linked with education, because
this is something I have predict

790
00:30:42,360 --> 00:30:45,360
I have thought would happen for
a long time and it is happening

791
00:30:45,360 --> 00:30:50,080
now is AI tutors this this idea
of education because I think we

792
00:30:50,080 --> 00:30:52,400
see education in general is
quite safe from the AI

793
00:30:52,400 --> 00:30:55,920
revolution in that whether we
like it or not we are to a

794
00:30:55,920 --> 00:30:59,560
certain degree childcare for the
for the society to go out and do

795
00:30:59,560 --> 00:31:01,080
work.
Yeah, children need somewhere to

796
00:31:01,080 --> 00:31:02,240
go during.
Safety in that.

797
00:31:02,320 --> 00:31:05,000
And it's also in terms of, you
know, education and school.

798
00:31:05,200 --> 00:31:07,520
I'm not trying to diminish the
job we do in terms of education,

799
00:31:07,520 --> 00:31:10,040
but I think it would be quite
daft for us to pretend that

800
00:31:10,040 --> 00:31:12,760
we're not also serving the
purpose of looking after

801
00:31:12,760 --> 00:31:14,240
children in our society so it
can function.

802
00:31:14,480 --> 00:31:16,400
Yeah, so.
So that's always where I think

803
00:31:16,400 --> 00:31:19,120
we've got kind of a safe
safeness to our job because

804
00:31:19,320 --> 00:31:20,640
children always need somewhere
to go.

805
00:31:20,840 --> 00:31:25,560
However, the actual idea of
tutoring to me, having seen this

806
00:31:25,560 --> 00:31:28,560
video of an AI bot talking to a
child and responding and

807
00:31:28,560 --> 00:31:31,920
teaching them something, picking
up misconceptions, giving them

808
00:31:31,920 --> 00:31:34,800
more, more, more questions of
what they need to practise and

809
00:31:34,800 --> 00:31:37,600
stuff that is real.
That that when people say, oh,

810
00:31:37,760 --> 00:31:40,320
you can never really build a
bond with a child or oh, you can

811
00:31:40,320 --> 00:31:42,920
never really understand the
misconceptions of a child, yes,

812
00:31:42,920 --> 00:31:44,680
you could.
Because we are at the point now

813
00:31:44,680 --> 00:31:47,800
where it is getting so close to
what a human brain does in terms

814
00:31:47,800 --> 00:31:50,880
of reasoning and problem solving
that you need to just see that

815
00:31:50,880 --> 00:31:53,920
video search AI Tutor now and
see it happen in real time with

816
00:31:53,920 --> 00:31:56,320
children.
It is incredibly good.

817
00:31:56,320 --> 00:31:57,160
Yeah.
And is it perfect?

818
00:31:57,160 --> 00:31:59,840
No, But I tell you what, it will
get good extremely quickly.

819
00:31:59,960 --> 00:32:01,360
The trajectory of AI is, is
crazy.

820
00:32:01,360 --> 00:32:03,440
And the last thing I guess I'll
mention on it is you probably

821
00:32:03,440 --> 00:32:05,600
notice if you're on social media
a lot, you'll notice that

822
00:32:05,600 --> 00:32:07,200
there's a lot of AI video
generation now.

823
00:32:07,200 --> 00:32:09,480
So we kind of had the whole
image generation revolution like

824
00:32:09,720 --> 00:32:11,560
not even a year ago.
That's how fast this is

825
00:32:11,560 --> 00:32:13,360
developing, right?
And they just were bad.

826
00:32:13,360 --> 00:32:13,960
Weren't they?
Yeah, they were.

827
00:32:13,960 --> 00:32:15,400
They were terrible at 1st, and
so was video.

828
00:32:15,400 --> 00:32:16,800
A couple of years ago it was
laughable.

829
00:32:16,800 --> 00:32:19,240
And even honestly, six months
ago, AI video was laughable.

830
00:32:19,480 --> 00:32:21,680
It's at the point now where it's
basically indistinguishable from

831
00:32:21,680 --> 00:32:23,320
real life.
It's completely insanely good.

832
00:32:23,320 --> 00:32:25,040
It has audio built in as as
well.

833
00:32:25,040 --> 00:32:27,800
So you've got voices, you've got
music, soundtracks and visuals

834
00:32:27,800 --> 00:32:30,480
now completely being generated
from written prompts, have AI to

835
00:32:30,480 --> 00:32:32,440
the point where I've even seen,
I haven't told you this before,

836
00:32:32,440 --> 00:32:35,320
but I've even seen now people
are developing apps for just

837
00:32:35,320 --> 00:32:37,720
making podcasts and you
basically just say what I want

838
00:32:37,720 --> 00:32:39,600
the podcast to be, what I want
the host to look like, what I

839
00:32:39,600 --> 00:32:41,800
want them to talk about the
length, press the button, there

840
00:32:41,800 --> 00:32:43,120
you go.
Just generates that and that's

841
00:32:43,120 --> 00:32:43,640
what we are.
Yeah.

842
00:32:43,640 --> 00:32:46,160
Can you can you generate one
about writing moderation that

843
00:32:46,160 --> 00:32:49,480
leads nicely into writing about
AI and where AI is self kind of

844
00:32:49,520 --> 00:32:52,400
looking at itself and and just
assessing where it's going?

845
00:32:52,400 --> 00:32:54,600
To go in the future and you can.
And we we laugh.

846
00:32:54,840 --> 00:32:56,600
Honestly, we could make one
that's probably relatively

847
00:32:56,600 --> 00:32:58,040
indistinguishable from what
we're producing.

848
00:32:58,080 --> 00:33:00,200
Yeah, 100%.
I know you heard of the podcast

849
00:33:00,200 --> 00:33:03,400
Diver CEO Stephen Bartlett.
He he said he's got to the point

850
00:33:03,400 --> 00:33:07,720
now where he's created like a, a
sister parallel podcast where

851
00:33:07,720 --> 00:33:09,200
his own voice has been trained
on it.

852
00:33:09,200 --> 00:33:14,840
So it's him speaking which gets
as much click through and user

853
00:33:14,840 --> 00:33:17,960
retention for an episode as his
actual one does.

854
00:33:18,040 --> 00:33:20,080
Wow.
So, so the, the, the, the, it's

855
00:33:20,080 --> 00:33:21,160
already there.
It already exists.

856
00:33:21,160 --> 00:33:23,680
It's just so you know, it's,
it's not like, oh, you click on

857
00:33:23,680 --> 00:33:24,720
it.
This is a fun gimmick.

858
00:33:24,920 --> 00:33:26,920
Oh, I'm listening for for two
minutes and I'll click off it.

859
00:33:27,160 --> 00:33:30,200
He's getting the same retention
rate on episodes because it's

860
00:33:30,200 --> 00:33:31,280
indistinguishable.
Yeah.

861
00:33:31,280 --> 00:33:34,760
So I guess my wider point here
about AI is, look, this is

862
00:33:34,760 --> 00:33:36,880
happening.
We we can have this discussion

863
00:33:36,880 --> 00:33:39,400
about whether it's safe, where
you can have this discussion

864
00:33:39,400 --> 00:33:41,120
about whether it's good for the
environment that we'll come on

865
00:33:41,120 --> 00:33:42,280
to later on.
We can have we can have all

866
00:33:42,280 --> 00:33:45,240
these discussions and that's
lovely and it's fantastic.

867
00:33:45,360 --> 00:33:46,640
But it's happening whether you
like it.

868
00:33:46,640 --> 00:33:48,480
Or not.
Oh yes, it's already way too far

869
00:33:48,480 --> 00:33:49,120
gone.
It's too far.

870
00:33:49,120 --> 00:33:50,440
It's there.
It's basically a race.

871
00:33:50,440 --> 00:33:53,360
It's like the best way I sort of
described was by Sam Altman who

872
00:33:53,360 --> 00:33:56,600
made Open AI which made chat and
he was saying it was like a bit

873
00:33:56,600 --> 00:33:58,400
like the moon race.
Firstly, it's just it's just

874
00:33:58,400 --> 00:34:00,560
happening and the and it's not
the race to the moon now, it's

875
00:34:00,560 --> 00:34:02,680
race to super intelligence.
He was like, I will be the first

876
00:34:02,680 --> 00:34:04,080
person to create super
intelligence.

877
00:34:04,360 --> 00:34:06,480
I will make an AI that's clever
than all human beings put

878
00:34:06,480 --> 00:34:08,480
together.
And the current trajectory of

879
00:34:08,480 --> 00:34:09,719
that is within five years.
Yeah.

880
00:34:09,880 --> 00:34:11,679
So it is like.
I think I think the scales are

881
00:34:11,679 --> 00:34:13,120
low.
It is mad and that scale has

882
00:34:13,120 --> 00:34:14,920
gone down, down, down.
You know, three years ago people

883
00:34:14,920 --> 00:34:18,920
were like, oh, by 2050, by 2029,
probably realistically we'll

884
00:34:18,920 --> 00:34:22,400
have probably have a super
intelligent AI that can just do

885
00:34:22,400 --> 00:34:24,719
absolutely everything and think
like us and utilise.

886
00:34:24,960 --> 00:34:26,920
I just think it's it's what will
happen.

887
00:34:27,040 --> 00:34:29,280
I just do.
So let's let's reframe this in

888
00:34:29,280 --> 00:34:31,960
terms of education now.
Yeah, because OK, again, we can

889
00:34:31,960 --> 00:34:34,040
have these long term discussions
about where I was going.

890
00:34:34,159 --> 00:34:36,400
Let's park that for a minute.
Make existential thread.

891
00:34:36,400 --> 00:34:39,159
Yeah, over there.
I'm just going to put it into my

892
00:34:39,159 --> 00:34:41,239
nice little packet and just
ignore that for a second.

893
00:34:41,360 --> 00:34:42,480
We'll come back to that later
on.

894
00:34:43,120 --> 00:34:46,320
But in, in terms of AI for
education, there are lots of

895
00:34:46,320 --> 00:34:48,480
good use cases.
There are lots of fantastic ways

896
00:34:48,480 --> 00:34:51,520
we can use AI to help.
And I think the traditional way

897
00:34:51,520 --> 00:34:54,600
of they're called LLM you, you
mentioned earlier large language

898
00:34:54,600 --> 00:34:56,679
models, which is what ChatGPT
essentially is.

899
00:34:56,960 --> 00:34:59,440
It has access to the entirety of
the Internet, which is huge.

900
00:34:59,720 --> 00:35:02,880
And then based on what you put
it predicts what a good response

901
00:35:02,880 --> 00:35:05,200
would be essentially, and it's
very good at doing that.

902
00:35:05,200 --> 00:35:07,880
So you can create texts very
quickly.

903
00:35:07,960 --> 00:35:10,640
You know, I've created diary
entries, I've created newspaper

904
00:35:10,640 --> 00:35:14,040
reports, I've created setting
descriptions, and they can all

905
00:35:14,040 --> 00:35:17,480
do that based on prompts that I
give them and create texts that

906
00:35:17,480 --> 00:35:20,440
my children can use.
And I can edit myself and even

907
00:35:20,440 --> 00:35:23,400
some questions, have some
retrieval questions, have some

908
00:35:23,520 --> 00:35:26,520
inference questions, find the
five trickiest words in here and

909
00:35:26,520 --> 00:35:28,840
do some word meaning questions
in this style, etcetera.

910
00:35:29,200 --> 00:35:30,960
It's really, really good at
that.

911
00:35:31,080 --> 00:35:34,080
And I think in general, a lot of
people kind of understand that

912
00:35:34,600 --> 00:35:36,680
and.
Why that can be useful reports,

913
00:35:36,680 --> 00:35:37,920
we'll come into more detail in a
second.

914
00:35:37,920 --> 00:35:40,440
Then we'll try and try and
explain to you guys how you can

915
00:35:40,440 --> 00:35:41,640
actually use that to help your
reports.

916
00:35:41,640 --> 00:35:44,080
But what, what else can AI do at
the minute?

917
00:35:44,520 --> 00:35:46,640
Well, firstly, I just do a
little bit more on the text

918
00:35:46,640 --> 00:35:48,600
thing, right?
One really good thing to

919
00:35:48,600 --> 00:35:51,520
understand is even when you're
generating text, it's so much

920
00:35:51,520 --> 00:35:53,320
better now than it was even a
year ago when we talked about

921
00:35:53,320 --> 00:35:54,800
this.
You can really, really tailor

922
00:35:54,800 --> 00:35:57,280
the style of it as well.
Like if you wanted to write it

923
00:35:57,280 --> 00:36:00,440
with a particular flow or like
look a particular way with its

924
00:36:00,440 --> 00:36:02,520
grammar and punctuation, you can
really just prompt it to do that

925
00:36:02,520 --> 00:36:04,080
now.
So don't, don't assume that, oh,

926
00:36:04,080 --> 00:36:06,720
everything looks like AI.
It kind of does unless you give

927
00:36:06,720 --> 00:36:09,080
it a bit more prompting to, to
do to create the text exactly

928
00:36:09,080 --> 00:36:10,240
how you want.
So you think, I think that's a

929
00:36:10,240 --> 00:36:11,320
good point.
It's as good as the prompt,

930
00:36:11,320 --> 00:36:12,200
isn't it?
Yeah, yeah, yeah.

931
00:36:12,240 --> 00:36:14,200
And again, we're going to kind
of crossover with that when we

932
00:36:14,200 --> 00:36:16,160
talk about reports again, but
that's one thing.

933
00:36:16,160 --> 00:36:18,320
But you can also, I mean, word
banks are another great one just

934
00:36:18,320 --> 00:36:19,560
for text.
I use it all the time for word

935
00:36:19,560 --> 00:36:21,240
banks, especially on the spot
where I've not thought about

936
00:36:21,240 --> 00:36:22,120
something.
I'm like, oh, do you know what?

937
00:36:22,120 --> 00:36:24,960
The kids really just, they keep
asking me, what's another word

938
00:36:24,960 --> 00:36:25,760
for this?
Sure.

939
00:36:26,440 --> 00:36:28,080
And I'm like, sometimes I want
to say I go and get a

940
00:36:28,080 --> 00:36:30,400
dictionary, but you know what?
In the modern day, honestly,

941
00:36:30,520 --> 00:36:33,360
honestly, it's a waste of time.
I'd rather I'd rather they had

942
00:36:33,360 --> 00:36:34,400
they don't because there's no
money.

943
00:36:34,600 --> 00:36:36,720
I'd rather they had a little,
little little app with them

944
00:36:36,720 --> 00:36:38,680
where they can just Google it or
synonyms for this word.

945
00:36:38,680 --> 00:36:39,920
You know what?
That's what I'd do in real life.

946
00:36:40,200 --> 00:36:42,080
So what it often ends up
happening in the middle ground

947
00:36:42,080 --> 00:36:45,160
here is I'll very quickly whip
up a a word Bank of things that

948
00:36:45,160 --> 00:36:46,960
were suitable for that genre,
including what they want.

949
00:36:46,960 --> 00:36:48,400
And then be like, OK, right,
We've got a word bank here,

950
00:36:48,400 --> 00:36:49,680
guys.
If you're really stuck for words

951
00:36:49,680 --> 00:36:52,280
for said look, here's 20 more or
you know, rather than asking me

952
00:36:52,280 --> 00:36:54,360
every single time, just stuff
like that is really good, but

953
00:36:54,360 --> 00:36:56,400
also image generation.
I know you used it, you've

954
00:36:56,400 --> 00:36:58,480
mentioned it on the pod before.
How you you got their

955
00:36:58,480 --> 00:36:59,320
descriptions?
So.

956
00:36:59,320 --> 00:37:00,440
Engaging.
Into it, right?

957
00:37:00,800 --> 00:37:02,440
Like image generation is
insanely good.

958
00:37:02,440 --> 00:37:05,480
Now you'll do honestly, anything
you write it to do, it'll be.

959
00:37:05,480 --> 00:37:07,560
It'll be pretty good at it.
I used it the other day.

960
00:37:07,600 --> 00:37:09,200
We know literally recently we
did.

961
00:37:09,400 --> 00:37:11,240
We did a setting.
We've been reading Narnia, and

962
00:37:11,240 --> 00:37:13,120
we did a setting description.
I can't remember if I mentioned

963
00:37:13,120 --> 00:37:15,920
this in the podcast or not, but
essentially we did some writing

964
00:37:15,920 --> 00:37:19,080
where we took the kind of
structure of Narnia in terms of

965
00:37:19,320 --> 00:37:23,120
a child walking through a
portal, which wardrobe is.

966
00:37:23,520 --> 00:37:25,920
So we took it as, OK, look,
we're going to do it writing.

967
00:37:26,120 --> 00:37:28,120
You are yourself.
You're writing a diary entry of

968
00:37:28,120 --> 00:37:32,200
your day and it started off of
you walking through whatever you

969
00:37:32,200 --> 00:37:33,880
want to walk through.
Some children chose a Bush.

970
00:37:33,880 --> 00:37:35,320
That one child had a really good
idea.

971
00:37:35,320 --> 00:37:37,560
They were playing football.
The football got kicked into the

972
00:37:37,560 --> 00:37:39,360
Bush.
They went through the Bush to

973
00:37:39,360 --> 00:37:41,440
find football, but they kept
going and kept going and then

974
00:37:41,440 --> 00:37:43,120
eventually got the ball, but
they were in a desert.

975
00:37:43,400 --> 00:37:45,160
Landscape.
So you can change the landscape

976
00:37:45,160 --> 00:37:47,440
at the end, you can change how
what what door you go through.

977
00:37:47,440 --> 00:37:48,680
You can change how you got
there.

978
00:37:48,880 --> 00:37:50,800
You can change whether you were
your siblings or your friends,

979
00:37:50,800 --> 00:37:51,960
how you felt about the
experience.

980
00:37:51,960 --> 00:37:54,480
Exactly that.
But what ChatGPT did amazingly

981
00:37:54,480 --> 00:37:57,160
well then was I said you need to
be really vivid with your

982
00:37:57,160 --> 00:37:59,640
descriptions.
Then it gave them that purpose

983
00:37:59,840 --> 00:38:03,440
because I said I'm going to feed
this into the AI for it to

984
00:38:03,480 --> 00:38:07,840
create a picture for you and it
will only go off your prompt.

985
00:38:08,160 --> 00:38:11,320
So if you're vague, it will give
you something that you're

986
00:38:11,400 --> 00:38:13,760
probably not imagining in your.
Head, you've got to be clear if

987
00:38:13,760 --> 00:38:15,200
you're.
Specific and clearer and use the

988
00:38:15,200 --> 00:38:17,520
right language, the better the
picture will be.

989
00:38:17,840 --> 00:38:19,520
I've never seen.
So much description.

990
00:38:19,520 --> 00:38:22,280
And then I picked about two or
three of what I deemed the

991
00:38:22,280 --> 00:38:26,160
children who tried to do that
the most, whacked it into AI and

992
00:38:26,160 --> 00:38:29,280
it made these images.
And it was 1 was stepping

993
00:38:29,280 --> 00:38:31,600
through, 1 was stepping through
the Bush.

994
00:38:31,640 --> 00:38:35,200
Like I mentioned, 1 was like the
door of a rocket, Right.

995
00:38:35,240 --> 00:38:36,960
So like, they go, they go into a
rocket.

996
00:38:37,160 --> 00:38:38,880
They, they go on their journey,
they step out.

997
00:38:38,880 --> 00:38:40,880
Yeah, they're in a new land.
You could have travelled there,

998
00:38:41,080 --> 00:38:43,080
but it's the idea of you're
still stepping through into this

999
00:38:43,080 --> 00:38:43,960
new land.
Yeah, yeah, yeah.

1000
00:38:43,960 --> 00:38:45,760
And AI did that and it was in
space.

1001
00:38:45,880 --> 00:38:50,280
And this other, this other boy
I've net, it's a classic boy

1002
00:38:50,280 --> 00:38:52,920
where you think it's, it can be
like drawing blood out of a

1003
00:38:52,920 --> 00:38:54,840
stone, trying lots of
description in their writing

1004
00:38:54,840 --> 00:38:57,760
because he's just not bothered.
And it's fine that he's not

1005
00:38:57,760 --> 00:38:59,520
bothered because he's 8.
And why would he care about this

1006
00:38:59,520 --> 00:39:01,000
thing?
He's got something else he wants

1007
00:39:01,000 --> 00:39:03,360
to think about.
But as soon as he realised that

1008
00:39:03,360 --> 00:39:06,720
it was going to be created into
an actual picture based on his

1009
00:39:06,720 --> 00:39:09,720
words and his words would come
to life immediately as well.

1010
00:39:09,720 --> 00:39:13,280
Immediately that that was enough
for him to be like, I want to be

1011
00:39:13,280 --> 00:39:15,080
picked, I'm gonna do this.
So I chose him at the end.

1012
00:39:15,360 --> 00:39:17,680
And his whole idea was the world
was upside down.

1013
00:39:17,800 --> 00:39:21,560
He stepped into an upside with
trees growing down and what was

1014
00:39:21,560 --> 00:39:23,880
really amazing, he thought of
all this himself in his world.

1015
00:39:23,880 --> 00:39:26,360
The rain went up okay, because
obviously it's upside down.

1016
00:39:26,720 --> 00:39:28,960
So there was rain going up from
the floor, down from the trees.

1017
00:39:28,960 --> 00:39:31,360
There were some buildings and
houses.

1018
00:39:31,760 --> 00:39:33,840
He wrote this really interesting
and he said what colour was it

1019
00:39:33,840 --> 00:39:37,840
was dark and gloomy and purpley
and, and it made it and his like

1020
00:39:37,840 --> 00:39:40,520
his face lit up.
And that's to me, that's the

1021
00:39:40,520 --> 00:39:42,960
power of this kind of thing
where it can give purpose to

1022
00:39:42,960 --> 00:39:45,960
some children who didn't have it
before where me saying, oh guys,

1023
00:39:46,080 --> 00:39:47,960
it's for imagination isn't that
great?

1024
00:39:47,960 --> 00:39:49,520
And and you the words can come
to life.

1025
00:39:49,680 --> 00:39:52,640
For some kids that's not enough.
They don't have that ability to

1026
00:39:52,640 --> 00:39:55,960
imagine that kind of thing.
But when the when the computer

1027
00:39:55,960 --> 00:39:59,440
can make it in 1.5 seconds.
Yeah, I got better writing out

1028
00:39:59,440 --> 00:40:01,040
of this child and there's so
much potential after that.

1029
00:40:01,040 --> 00:40:03,120
You can do all sorts of things
that say it doesn't generate

1030
00:40:03,120 --> 00:40:05,080
what they want.
OK, Analyse that picture.

1031
00:40:05,080 --> 00:40:07,160
What's it missing?
Where can you edit that in your

1032
00:40:07,160 --> 00:40:08,520
writing?
Go back what is clearly you've

1033
00:40:08,520 --> 00:40:10,120
got something in your head that
you've not described clearly.

1034
00:40:10,120 --> 00:40:11,080
There's so many things you can
do.

1035
00:40:11,400 --> 00:40:14,320
A really interesting parallel
came to my mind then when you

1036
00:40:14,320 --> 00:40:16,560
when you were talking about this
and it and it is hilariously a

1037
00:40:16,560 --> 00:40:19,520
parallel because 10 years ago,
and it wasn't my first year of

1038
00:40:19,520 --> 00:40:20,680
teaching because I've been
teaching 10 years.

1039
00:40:20,680 --> 00:40:23,520
You've been teaching nine years,
not quite a decade like me, 10

1040
00:40:23,520 --> 00:40:26,720
years ago.
I also happened to be doing

1041
00:40:26,720 --> 00:40:30,360
Narnia in a different school.
And a task that I thought of and

1042
00:40:30,360 --> 00:40:31,960
I, and I was really proud of
this task because I thought this

1043
00:40:31,960 --> 00:40:36,680
is really cool was it was a
scene with Mr Tumnus in, in his,

1044
00:40:36,720 --> 00:40:38,880
in his house.
And Lucy was there and they were

1045
00:40:38,880 --> 00:40:42,560
doing some setting description.
And I said, I gave them this

1046
00:40:42,560 --> 00:40:44,720
colouring in sheet, which was a
picture of that.

1047
00:40:44,880 --> 00:40:47,480
And I said, you're only allowed
to colour bits in, in the

1048
00:40:47,480 --> 00:40:49,760
picture when you've written it
into one of your sensors.

1049
00:40:50,120 --> 00:40:52,240
And the, it was very much the
same concept of what you're

1050
00:40:52,240 --> 00:40:53,120
doing there.
I was like it.

1051
00:40:53,240 --> 00:40:55,000
I was thinking that it's going
to make them be vivid with their

1052
00:40:55,000 --> 00:40:56,760
descriptions because they can't
colour it in unless they've

1053
00:40:56,760 --> 00:40:58,680
written about it.
And that was the idea.

1054
00:40:58,680 --> 00:41:01,680
And I just love how what you've
done is, is the, is the natural

1055
00:41:01,680 --> 00:41:04,000
advancement of that using
technology to the absolute MTH

1056
00:41:04,000 --> 00:41:05,720
degree because they got instant
gratification.

1057
00:41:05,920 --> 00:41:07,840
Wildly different pictures.
Mine were all said they all had

1058
00:41:07,840 --> 00:41:10,120
the exact same colouring engine.
Given the freedom doesn't.

1059
00:41:10,120 --> 00:41:12,000
It they had so much more read
them, you could immediately

1060
00:41:12,000 --> 00:41:14,000
analyse it and change it, you
know, and you could retype bits

1061
00:41:14,000 --> 00:41:16,560
in that's yeah, it's so.
Interesting, what we're

1062
00:41:16,560 --> 00:41:18,320
basically saying is my lesson
was way better than yours.

1063
00:41:18,480 --> 00:41:20,800
Yeah, but ten years ago I was,
you know, well ahead of the

1064
00:41:20,800 --> 00:41:22,720
curve, mate.
I basically invented the idea.

1065
00:41:22,840 --> 00:41:24,880
But you, but that's not even a
bad lesson because what what

1066
00:41:24,880 --> 00:41:27,280
you're doing there, you're still
teaching children to take things

1067
00:41:27,280 --> 00:41:28,760
into account when you're doing a
setting description.

1068
00:41:28,760 --> 00:41:30,400
Don't get me wrong, I think
that's still a good thing.

1069
00:41:30,680 --> 00:41:33,960
But yeah, this just unlocked it
in terms of the the ceiling is

1070
00:41:33,960 --> 00:41:36,720
your imagination.
The ceiling is your imagination

1071
00:41:36,720 --> 00:41:38,000
now.
So the more you're descriptive,

1072
00:41:38,000 --> 00:41:40,000
the more you explain, the more
it will actually be in

1073
00:41:40,000 --> 00:41:42,480
existence.
And I just find personally that

1074
00:41:42,480 --> 00:41:47,360
that hook is so powerful for
writing to inspire children who

1075
00:41:47,360 --> 00:41:49,800
otherwise aren't inspired and
struggle to make that link

1076
00:41:49,800 --> 00:41:52,400
between what they can imagine
and what they can write down.

1077
00:41:52,440 --> 00:41:55,800
Yeah, this was like almost a
bridge between that, where I got

1078
00:41:55,800 --> 00:41:58,280
more interesting.
Look, did it always make sense?

1079
00:41:58,280 --> 00:41:59,680
Was it always perfectly
punctuated?

1080
00:41:59,840 --> 00:42:00,960
That's a separate part of my
job.

1081
00:42:00,960 --> 00:42:02,520
And that's fine.
But you know what?

1082
00:42:02,520 --> 00:42:05,520
Realistically, as a teacher, at
the end of the day, I was so

1083
00:42:05,520 --> 00:42:07,920
excited that he was excited to
write.

1084
00:42:07,920 --> 00:42:09,400
Yeah.
Did you make his own world up?

1085
00:42:09,480 --> 00:42:11,040
Yeah, it did.
It Did they aspire.

1086
00:42:11,120 --> 00:42:12,640
Have to write something yeah,
exactly.

1087
00:42:12,640 --> 00:42:14,080
Cool.
Done That's that's like 80% of

1088
00:42:14,080 --> 00:42:15,240
your job done.
That's it honestly, yeah,

1089
00:42:15,480 --> 00:42:18,000
honestly when you're when you're
14 do you know what if it

1090
00:42:18,000 --> 00:42:20,760
doesn't click by then then
you've still got time because

1091
00:42:20,760 --> 00:42:22,280
you've got a brain that's over
14 year old.

1092
00:42:22,280 --> 00:42:24,360
And I can say to you, by the
way, guy that needs to have a

1093
00:42:24,360 --> 00:42:26,360
that needs to have a full stop
there, that needs to have a

1094
00:42:26,400 --> 00:42:28,080
capital ethic, by the way, you'd
probably learn that before

1095
00:42:28,080 --> 00:42:30,000
you're an adult, right?
And it's like, yeah, sure, it's,

1096
00:42:30,040 --> 00:42:31,760
it's OK.
It's funny.

1097
00:42:31,880 --> 00:42:35,200
We have archaic systems of, of
grids of what we expect children

1098
00:42:35,200 --> 00:42:37,480
to be able to do by their eight.
Forget it.

1099
00:42:37,480 --> 00:42:39,640
I just don't care.
I'm going to teach them it.

1100
00:42:39,640 --> 00:42:42,200
I'm going to, I'm, I'm, I'm not
saying let's not do that.

1101
00:42:42,200 --> 00:42:43,800
I'm not saying it's not
important to teach them because

1102
00:42:43,800 --> 00:42:46,000
other children will pick it up.
But if some kids still don't

1103
00:42:46,000 --> 00:42:49,560
quite get it after the 57th time
of me saying it, it's OK if they

1104
00:42:49,560 --> 00:42:51,960
don't get it the 58th time.
But you know, what's amazing is

1105
00:42:51,960 --> 00:42:53,400
that child now wants to do some
writing.

1106
00:42:53,640 --> 00:42:56,400
Win, win, win.
And that's what I think my job

1107
00:42:56,400 --> 00:42:59,200
is.
The obsession with metrics has

1108
00:42:59,200 --> 00:43:01,120
really flipped what it's
supposed to be about.

1109
00:43:01,440 --> 00:43:04,680
The foundation in the culture of
teaching, I think, is the

1110
00:43:04,680 --> 00:43:06,760
foundation is we teach them how
to do all the grammar,

1111
00:43:06,760 --> 00:43:09,120
punctuation and spelling, and on
top of that they can then write

1112
00:43:09,120 --> 00:43:10,120
lovely things.
No, it's not.

1113
00:43:10,280 --> 00:43:13,080
The foundation should be where
kids want to write something

1114
00:43:13,560 --> 00:43:16,560
honestly, just if you've done
that should be the the absolute

1115
00:43:16,560 --> 00:43:19,280
first that you work on because
progress is fairly natural for

1116
00:43:19,280 --> 00:43:22,480
most children if they just want
to do it honestly.

1117
00:43:22,480 --> 00:43:25,560
Like if you, if a kid enjoys
writing for seven years in a

1118
00:43:25,560 --> 00:43:28,600
row, it's very, very, very
unlikely.

1119
00:43:28,800 --> 00:43:32,240
And there probably will be
reasons for it if not that they

1120
00:43:32,240 --> 00:43:35,320
won't make progress in, in the
stuff that we think is more

1121
00:43:35,320 --> 00:43:37,200
important than metrics, the
first stuff we can measure.

1122
00:43:37,400 --> 00:43:39,280
And I think if we had the
attitude as teachers, we would

1123
00:43:39,440 --> 00:43:42,360
be creating a better generation
of writers in general.

1124
00:43:42,360 --> 00:43:44,680
They might just not get some of
the stuff as early as we think.

1125
00:43:44,760 --> 00:43:46,440
Oh no.
But they like writing still when

1126
00:43:46,440 --> 00:43:49,280
they're 15, so.
As early as we've arbitrarily

1127
00:43:49,280 --> 00:43:52,760
decided they need to like who,
who actually decided that 8 year

1128
00:43:52,760 --> 00:43:54,960
olds need to use frontal verbals
in their text free like

1129
00:43:54,960 --> 00:43:56,240
frequently and different
phrases?

1130
00:43:56,800 --> 00:43:58,720
Like so it's not objective.
Is it Yeah.

1131
00:43:58,720 --> 00:44:00,360
It's not objective that they put
the comma there or not.

1132
00:44:00,680 --> 00:44:03,880
I, you know what really, I don't
really care too much.

1133
00:44:03,880 --> 00:44:07,200
And the great thing about AI and
other use of, of it in school is

1134
00:44:07,680 --> 00:44:09,920
it is getting better at doing
that anyway for teachers that it

1135
00:44:09,920 --> 00:44:12,440
can be, it can become an
analysing tool.

1136
00:44:12,520 --> 00:44:16,000
You can upload a picture of
something that someone's written

1137
00:44:16,360 --> 00:44:18,800
and you can ask it to based on
the year 4 curriculum.

1138
00:44:18,960 --> 00:44:21,800
Can you please give me some pros
and cons for this piece of

1139
00:44:21,800 --> 00:44:23,680
writing, Some strengths and some
weaknesses or some next steps,

1140
00:44:23,680 --> 00:44:26,240
etcetera.
And it will be able to very

1141
00:44:26,240 --> 00:44:28,440
quickly say, oh, we're using
sentences that makes sense.

1142
00:44:28,440 --> 00:44:30,960
That's awesome.
There are some fronted clauses

1143
00:44:30,960 --> 00:44:33,240
in there for maybe their
prepositions, maybe adverbs or

1144
00:44:33,240 --> 00:44:36,360
some whatever is going on.
The the child isn't always clear

1145
00:44:36,360 --> 00:44:38,120
of their writing, their
descendants, Those aren't going

1146
00:44:38,120 --> 00:44:40,480
down very far.
And also sometimes it just

1147
00:44:40,480 --> 00:44:42,200
doesn't quite make sense at the
end when they're trying to use

1148
00:44:42,200 --> 00:44:44,400
conjunctions.
Yeah.

1149
00:44:44,400 --> 00:44:46,360
OK, look, I can do that as an
experienced teacher quite

1150
00:44:46,360 --> 00:44:47,760
quickly.
But if you're learning your

1151
00:44:47,760 --> 00:44:50,520
trade, this is an amazing tool
to have next year as a teacher.

1152
00:44:50,640 --> 00:44:52,840
Or even if you're more.
Experienced, experienced, it's

1153
00:44:52,840 --> 00:44:55,440
just another pair of eyes.
I wouldn't say no to look at

1154
00:44:55,440 --> 00:44:57,400
myself and like, no, I can do
it, Dylan, I can do it.

1155
00:44:57,400 --> 00:44:59,320
But if you were offering to help
me like analyse looking through

1156
00:44:59,320 --> 00:45:00,800
some work and I'd be like, yeah,
cool, cheers.

1157
00:45:00,800 --> 00:45:04,800
Moderation, Why wouldn't I?
What moderation is AI is going

1158
00:45:04,800 --> 00:45:07,520
to be, and I'm saying this now
on the podcast because I think

1159
00:45:07,520 --> 00:45:09,600
in two years time when we're
still doing this podcast,

1160
00:45:09,600 --> 00:45:11,960
hopefully we'll look back and be
like, yeah, we were right.

1161
00:45:12,240 --> 00:45:15,640
AI is going to absolutely
transform the world in the next

1162
00:45:15,640 --> 00:45:17,040
three years.
It's really interesting because

1163
00:45:17,040 --> 00:45:18,840
that's what I've told you for a
year and you've told me no until

1164
00:45:18,840 --> 00:45:20,200
now.
And now you've stolen my

1165
00:45:20,200 --> 00:45:22,640
prediction.
So it was my, you heard it here

1166
00:45:22,640 --> 00:45:24,320
first from me.
You heard it here first.

1167
00:45:24,400 --> 00:45:27,440
Hayden from TCP, not Deeden,
that we talked about said that

1168
00:45:27,440 --> 00:45:28,440
AI is going to transform the
world.

1169
00:45:28,440 --> 00:45:29,760
But it really is.
And you're absolutely right.

1170
00:45:29,760 --> 00:45:31,360
Like I thought the trajectory
was much slower.

1171
00:45:31,600 --> 00:45:34,440
And I've only changed my mind
honestly in the last like two or

1172
00:45:34,440 --> 00:45:36,880
three months where I've seen
what AI was doing at the

1173
00:45:36,880 --> 00:45:39,800
exponential growth rate.
And I'm like, this is, this is

1174
00:45:39,800 --> 00:45:41,960
actually mad because we weren't
a year ago, we weren't supposed

1175
00:45:41,960 --> 00:45:44,680
to be doing these things.
And by this time it's it's just

1176
00:45:44,680 --> 00:45:46,840
completely crazy.
I think, yeah, in general,

1177
00:45:46,840 --> 00:45:47,960
there's always going to be some
gimmicks.

1178
00:45:47,960 --> 00:45:51,440
Like we said, we went to bet,
you know, 90% of the businesses

1179
00:45:51,440 --> 00:45:53,200
there had AI in the name and
stuff.

1180
00:45:53,600 --> 00:45:55,720
There's some, don't get me
wrong, there's some cool stuff

1181
00:45:55,720 --> 00:45:57,040
out there.
I think there's going to be a

1182
00:45:57,040 --> 00:46:00,000
split between what we can use
almost on a daily basis, which

1183
00:46:00,000 --> 00:46:02,200
is really having a massive
impact on the children.

1184
00:46:02,320 --> 00:46:04,280
Yeah.
And I, I, I feel rude calling

1185
00:46:04,280 --> 00:46:06,760
them gimmicks, but they kind of
are gimmicky to a degree because

1186
00:46:06,760 --> 00:46:08,640
they have such a limited use
niche tools.

1187
00:46:08,640 --> 00:46:10,160
Yeah.
Like, you know, making up,

1188
00:46:10,520 --> 00:46:12,920
inventing songs.
I think that's, that's an

1189
00:46:12,920 --> 00:46:14,960
amazing technology to be able to
create a song.

1190
00:46:15,080 --> 00:46:16,800
But in, in a school, I'll use
that.

1191
00:46:16,920 --> 00:46:19,200
Sometimes I'll make a song.
I've, I've spoke about this

1192
00:46:19,200 --> 00:46:21,960
before, I made a song before
about the water cycle because

1193
00:46:21,960 --> 00:46:23,600
that's something tricky to
remember.

1194
00:46:23,720 --> 00:46:25,120
It's another hook for some
children.

1195
00:46:25,120 --> 00:46:27,680
They might remember the song,
remember the Jingle, the words

1196
00:46:27,680 --> 00:46:29,360
are embedded.
We use the vocab, we do some

1197
00:46:29,360 --> 00:46:30,000
signs.
Yeah.

1198
00:46:30,000 --> 00:46:32,280
That's just pedagogy to me.
And I think there's, there's

1199
00:46:32,280 --> 00:46:34,360
always going to be those kind of
tools where we think as a

1200
00:46:34,360 --> 00:46:37,800
teacher, that's helpful for me
to teach this just like I used

1201
00:46:37,800 --> 00:46:39,080
to.
Maybe going to YouTube and find

1202
00:46:39,080 --> 00:46:40,200
a water cycle.
Yeah, yeah, yeah.

1203
00:46:40,200 --> 00:46:42,560
But I would spend ages trying to
find the right one and maybe I'd

1204
00:46:42,560 --> 00:46:44,600
pause it at one minute 47
because then it goes into

1205
00:46:44,600 --> 00:46:45,680
something that's irrelevant.
Yeah.

1206
00:46:45,840 --> 00:46:48,280
With this kind of AI tool, it
could be bespoke to you.

1207
00:46:48,280 --> 00:46:49,800
You can create it straight away
again.

1208
00:46:49,920 --> 00:46:51,880
You know, this, this is less
gimmicky, I guess, but in terms

1209
00:46:51,880 --> 00:46:53,640
of just on the idea of like,
they're just going to be more

1210
00:46:53,640 --> 00:46:55,440
and more and more tools and
tools and tools and tools and

1211
00:46:55,440 --> 00:46:57,400
tools as AI gets better.
And people just put wrappers

1212
00:46:57,400 --> 00:46:59,120
around them and say, here's a
really good tool we've packaged

1213
00:46:59,120 --> 00:47:00,960
up from it.
I saw something the other day

1214
00:47:00,960 --> 00:47:02,640
and I can't think of the name of
it, annoyingly, because it was

1215
00:47:02,640 --> 00:47:04,600
just sort of on my Instagram and
I and I watched the video and it

1216
00:47:04,600 --> 00:47:07,080
was insanely good at making
PowerPoints.

1217
00:47:07,480 --> 00:47:09,240
And I was like, I know that's
been around for a little while.

1218
00:47:09,240 --> 00:47:10,680
AI has kind of been involved in
like helping you make

1219
00:47:10,680 --> 00:47:12,320
PowerPoints, but this was like
next level.

1220
00:47:12,440 --> 00:47:15,960
It was basically just put in one
text prompt and it will use the

1221
00:47:15,960 --> 00:47:18,360
powers of the Internet and its
image and video creation to make

1222
00:47:18,360 --> 00:47:20,960
the most insanely good
PowerPoint presentation at

1223
00:47:20,960 --> 00:47:23,360
whatever level you want it for
either a lesson or for like a

1224
00:47:23,360 --> 00:47:27,080
business meeting, whatever it is
insanely good engaging

1225
00:47:27,080 --> 00:47:28,720
PowerPoint presentation on that
thing.

1226
00:47:28,800 --> 00:47:30,440
And I was like, do you know
what?

1227
00:47:30,640 --> 00:47:32,760
There are some teachers.
There are a lot of schools still

1228
00:47:32,760 --> 00:47:35,560
where maybe their schemes of
learning are not in place or

1229
00:47:35,560 --> 00:47:38,760
they change them every year
because they're toxic or, you

1230
00:47:38,760 --> 00:47:40,760
know, for whatever reason,
teachers are having to still

1231
00:47:40,760 --> 00:47:43,320
generate stuff that's been in
the career for 10 years.

1232
00:47:43,320 --> 00:47:44,640
It's been the same objective for
10 years.

1233
00:47:44,640 --> 00:47:46,880
We're still having to, for some
reason, generate new PowerPoints

1234
00:47:46,880 --> 00:47:48,320
for it.
And I was like that would be a

1235
00:47:48,320 --> 00:47:50,400
useful tool for them because
that will do that in 20 seconds.

1236
00:47:50,840 --> 00:47:52,560
Can you put the school logo in
the corner please?

1237
00:47:52,560 --> 00:47:54,160
And can you change the font to
Sansferi?

1238
00:47:54,160 --> 00:47:56,360
Serif, the AI has been trained
to say no.

1239
00:47:56,360 --> 00:47:57,680
Actually.
That's so ridiculous even I

1240
00:47:57,680 --> 00:48:00,400
won't do it.
You have to take yourself anyway

1241
00:48:00,400 --> 00:48:02,440
yourself.
Yeah, the AI knows how stupid

1242
00:48:02,440 --> 00:48:03,200
that is.
Right.

1243
00:48:03,200 --> 00:48:05,200
So in general, obviously it's
good use for AI.

1244
00:48:05,200 --> 00:48:07,920
We talked about our existential
dread about where the whole

1245
00:48:07,920 --> 00:48:08,800
planet's going.
So that's.

1246
00:48:08,800 --> 00:48:10,000
Been fun and the episode's over,
no?

1247
00:48:10,360 --> 00:48:13,080
No, no, no, because now we're
going to talk about specifically

1248
00:48:13,080 --> 00:48:15,400
for writing reports, because
Hayden and I have used AI to

1249
00:48:15,400 --> 00:48:17,440
help us write reports.
This will be the third year I

1250
00:48:17,440 --> 00:48:19,920
think you've used it.
I think it's getting better and

1251
00:48:19,920 --> 00:48:22,840
better, but we still have many,
many colleagues out there.

1252
00:48:23,040 --> 00:48:25,120
And we'll talk about some of the
attracting statements later,

1253
00:48:25,120 --> 00:48:27,760
maybe about why people argue
against doing this and why we

1254
00:48:27,760 --> 00:48:30,760
think it's OK.
But in general, Hayden, can you

1255
00:48:30,760 --> 00:48:35,760
just outline what does someone
have to do in order to save time

1256
00:48:35,960 --> 00:48:39,080
right in their reports using AI?
What do you even do?

1257
00:48:39,080 --> 00:48:40,720
What are the options?
What does it look like?

1258
00:48:40,720 --> 00:48:42,760
OK, I'll start about 400
questions.

1259
00:48:42,920 --> 00:48:44,840
I'll start with the options.
Firstly, there's there's free

1260
00:48:44,840 --> 00:48:47,520
and there's paid options.
By free I mean ChatGPT or any

1261
00:48:47,520 --> 00:48:50,080
other large language model.
I only use ChatGPT as anyone I

1262
00:48:50,080 --> 00:48:52,080
can recommend, right?
There are other paid services

1263
00:48:52,080 --> 00:48:54,520
out there.
I know that I see Mr P's got his

1264
00:48:54,840 --> 00:48:56,800
teach my AI.
I know there's another friend of

1265
00:48:56,800 --> 00:48:58,480
ours has made something called
Rocket AI and there's probably

1266
00:48:58,520 --> 00:49:00,520
other companies things out there
as well that do the same sort of

1267
00:49:00,520 --> 00:49:01,640
thing.
Really what they're doing is

1268
00:49:01,640 --> 00:49:03,840
they're putting a wrapper around
different language models with

1269
00:49:04,040 --> 00:49:05,600
lots of other other restrictions
and things in place.

1270
00:49:05,600 --> 00:49:07,640
I'm not going to pretend I know
exactly how they do it and

1271
00:49:07,640 --> 00:49:10,320
scaffolds scaffolds, but they
are essentially utilising large

1272
00:49:10,320 --> 00:49:12,640
language models and just
packaging it up to be a slightly

1273
00:49:12,640 --> 00:49:14,360
more easier.
I've used these models before

1274
00:49:14,360 --> 00:49:15,760
and they are really good because
you know what they do?

1275
00:49:15,840 --> 00:49:17,440
They package it up and make it
really easy.

1276
00:49:17,440 --> 00:49:20,440
I can put the kids name in, I
can put this what, what year

1277
00:49:20,440 --> 00:49:21,840
group they're in, what I want it
to be about.

1278
00:49:21,840 --> 00:49:24,040
I can put the formality, I can
bullet point some things in

1279
00:49:24,040 --> 00:49:25,640
there and it generates me a
lovely comment.

1280
00:49:25,800 --> 00:49:28,320
You can also do this on ChatGPT
if you get your prompts right.

1281
00:49:28,480 --> 00:49:31,120
So and.
ChatGPT has a free account

1282
00:49:31,120 --> 00:49:33,160
option right?
You can pay for ChatGPT and it

1283
00:49:33,160 --> 00:49:35,680
like it boosts how many models
you can get or how quick that

1284
00:49:35,680 --> 00:49:38,080
you get stuff.
But you can just sign up for

1285
00:49:38,080 --> 00:49:39,840
free to ChatGPT and get it to
do.

1286
00:49:39,840 --> 00:49:41,560
This yeah, you can and and I
wouldn't be surprised if you

1287
00:49:41,560 --> 00:49:43,520
know if you're listening to this
in a few months time or a year,

1288
00:49:43,520 --> 00:49:46,400
I wouldn't be surprised if this
information isn't perfectly

1289
00:49:46,400 --> 00:49:48,680
accurate because I know that
they are as they monetize it

1290
00:49:48,680 --> 00:49:50,240
more and as more people use it
that putting in more

1291
00:49:50,240 --> 00:49:52,680
restrictions for the free users.
They'd be like, oh, you've used

1292
00:49:52,680 --> 00:49:54,880
up your, you get to wait 10
minutes now to do another

1293
00:49:54,880 --> 00:49:56,360
prompt.
So like, don't you know, quote

1294
00:49:56,360 --> 00:49:58,000
me exactly on that because that
might change.

1295
00:49:58,000 --> 00:50:00,520
But we've used the paid, paid
version before and it was, it

1296
00:50:00,520 --> 00:50:01,640
was, it was very quick.
Yeah.

1297
00:50:02,400 --> 00:50:05,200
Anyway, so should we talk about
how to prompt now?

1298
00:50:05,200 --> 00:50:06,640
Should we talk?
About that, sure, because a lot

1299
00:50:06,640 --> 00:50:08,920
of people, again, you know, I've
comment, I've done memes and

1300
00:50:08,920 --> 00:50:13,800
stuff about I did one me which
said me, ChatGPT and I having a

1301
00:50:13,800 --> 00:50:16,120
dance after we finished our
reports and it's just me and

1302
00:50:16,120 --> 00:50:18,480
it's like it's a person and a
robot just doing a dance.

1303
00:50:18,480 --> 00:50:21,960
Like what do you mean dancing?
And you know, a lot of people go

1304
00:50:21,960 --> 00:50:24,120
haha, yeah, me too, ChatGPT is
with my mate, etcetera.

1305
00:50:24,240 --> 00:50:27,200
But we still get comments from
people saying how do I do this?

1306
00:50:27,200 --> 00:50:29,760
Yeah.
OK so here's my recommendation

1307
00:50:29,760 --> 00:50:31,800
because this is what I did this
year and this is the first time

1308
00:50:31,800 --> 00:50:33,760
I've done it this way by the way
and I thought it was extremely

1309
00:50:33,760 --> 00:50:35,520
effective.
And just for context to get you

1310
00:50:35,520 --> 00:50:38,880
excited, I did my reports in
under 2 hours this year and I'd

1311
00:50:38,880 --> 00:50:41,720
say they are by far the best
reports I've ever I will ever

1312
00:50:41,800 --> 00:50:45,440
deliver to parents and children.
One of the reasons is AI, as

1313
00:50:45,440 --> 00:50:48,360
we've said very clearly, it is
getting insanely better.

1314
00:50:48,360 --> 00:50:51,120
Like by the year.
If you had a year ago

1315
00:50:51,120 --> 00:50:54,200
experienced AI and thought now
this isn't for me, this isn't

1316
00:50:54,200 --> 00:50:56,480
quite how I want it.
Honestly, wipe that, give it

1317
00:50:56,480 --> 00:51:00,000
another go because it is tenfold
better now that doing its job.

1318
00:51:00,120 --> 00:51:01,440
So how so?
How do we do it?

1319
00:51:01,520 --> 00:51:03,800
So how do we do it?
So I did like this.

1320
00:51:03,800 --> 00:51:06,160
I had AI used the Word document,
right, because you want to know

1321
00:51:06,160 --> 00:51:08,560
you can, you can just drag word
documents into ChatGPT and or

1322
00:51:08,560 --> 00:51:10,440
just read the text on that.
You don't have to type directly

1323
00:51:10,440 --> 00:51:12,920
into it.
I, I put a made a Word document

1324
00:51:12,920 --> 00:51:15,400
and I made a table in it.
And on the left hand side, I

1325
00:51:15,400 --> 00:51:17,320
just listed the kids as ABCDEFG,
right.

1326
00:51:17,320 --> 00:51:19,120
So we're not going to be Oh, I
didn't want to use names for

1327
00:51:19,120 --> 00:51:21,040
obvious GDPR reasons.
I'm not going to get into it.

1328
00:51:21,040 --> 00:51:22,800
So don't use names.
I just put ABCD.

1329
00:51:22,800 --> 00:51:24,800
I know who they are because it's
just my, it's just my class

1330
00:51:24,800 --> 00:51:26,720
register in order.
So I know B is the second kid on

1331
00:51:26,720 --> 00:51:29,720
my register and then I put 2
columns to the right of it and

1332
00:51:29,720 --> 00:51:33,560
you can simplify it or
complexify it as much as you

1333
00:51:33,560 --> 00:51:35,560
want.
My first column was general

1334
00:51:35,560 --> 00:51:39,320
notes and I just bullet pointed
anything that came to mind based

1335
00:51:39,320 --> 00:51:42,440
on I know you did this as well,
based on like a little thing.

1336
00:51:42,440 --> 00:51:43,880
I Yeah.
Let's talk about that.

1337
00:51:43,880 --> 00:51:44,720
Yeah, we'll talk about.
In a second.

1338
00:51:44,720 --> 00:51:46,520
So yeah, you've done your
generic points, done my points.

1339
00:51:46,520 --> 00:51:48,480
And on the right hand side I put
specific points as well.

1340
00:51:48,480 --> 00:51:50,960
I put like actual like anecdotes
from the year that I would, I

1341
00:51:50,960 --> 00:51:52,080
would want to mention in the
report.

1342
00:51:52,080 --> 00:51:53,160
Not everyone got something in
that.

1343
00:51:53,360 --> 00:51:54,880
It depends if there was
something that came to mind.

1344
00:51:55,120 --> 00:51:57,960
I dragged that into ChatGPT and
I just wrote a prompt of it that

1345
00:51:57,960 --> 00:52:01,440
set the parameters and it was
along the lines of it's there

1346
00:52:01,440 --> 00:52:04,040
for end of year 6 reports in the
UKI want the word count to be

1347
00:52:04,040 --> 00:52:06,600
around 180 words.
I want, I want it to be

1348
00:52:06,600 --> 00:52:08,320
informal, but with a
professional tone.

1349
00:52:08,320 --> 00:52:09,560
You went informal?
Yeah.

1350
00:52:09,560 --> 00:52:11,960
I, I find when you say informal,
the professional tone, it

1351
00:52:12,000 --> 00:52:13,120
actually comes out way better
when you.

1352
00:52:13,120 --> 00:52:13,960
But it's formal.
Yeah.

1353
00:52:13,960 --> 00:52:15,680
It's still formal because it's
because it understands

1354
00:52:15,680 --> 00:52:18,240
professional, but it keeps it in
a tone where I would write,

1355
00:52:18,240 --> 00:52:18,760
Yeah.
Because.

1356
00:52:18,760 --> 00:52:20,880
Because reports have to be
professional still.

1357
00:52:20,920 --> 00:52:22,600
Yeah, but you want your voice to
come through.

1358
00:52:22,600 --> 00:52:24,080
Yeah, exactly.
So I found that as a little

1359
00:52:24,080 --> 00:52:26,000
cheat there, rather than putting
formal, I put informal, the

1360
00:52:26,000 --> 00:52:27,120
professional tone.
Brilliant.

1361
00:52:27,680 --> 00:52:29,920
A lot of other parameters of
exactly what I want, you know,

1362
00:52:29,920 --> 00:52:31,840
if I want to split into two
paragraphs, say split into two

1363
00:52:31,840 --> 00:52:35,640
paragraphs, if I said keep it
really positive, I want it to,

1364
00:52:35,640 --> 00:52:38,240
you know, I want it to be
friendly and approach well and

1365
00:52:38,240 --> 00:52:39,720
that's it.
Honestly, it's clever enough at

1366
00:52:39,720 --> 00:52:41,720
that point where it's like it
knows what year, it knows

1367
00:52:41,760 --> 00:52:44,280
honestly what end of year 6
reports are supposed to look

1368
00:52:44,280 --> 00:52:45,280
like.
For whatever reason.

1369
00:52:45,280 --> 00:52:47,640
It is good at that now.
It was right in things like, you

1370
00:52:47,640 --> 00:52:50,320
know, like referencing secondary
school.

1371
00:52:50,320 --> 00:52:51,720
I didn't tell it that we weren't
a free school.

1372
00:52:51,720 --> 00:52:54,240
It just knows that probably the
chances are if it's a year 6,

1373
00:52:54,240 --> 00:52:55,720
they're going to be going to a
secondary school next.

1374
00:52:55,720 --> 00:52:58,480
Year if you do all of it, like
you said, you you've done all of

1375
00:52:58,480 --> 00:52:59,880
that.
Maybe we'll come some push backs

1376
00:52:59,880 --> 00:53:02,520
in a second about it and and
things you can do to, you know,

1377
00:53:02,520 --> 00:53:04,160
streamline it.
You've done all that, but you're

1378
00:53:04,160 --> 00:53:06,680
still not happy with it.
What would what would you say to

1379
00:53:06,680 --> 00:53:08,120
someone?
Because you spent a lot of time

1380
00:53:08,120 --> 00:53:10,240
making that table and I can
imagine if people didn't get

1381
00:53:10,240 --> 00:53:11,720
what they want straight away
them I think what a waste of

1382
00:53:11,720 --> 00:53:13,440
time this is.
This is silly and one of the

1383
00:53:13,440 --> 00:53:15,680
push backs I see is it takes
just as long as me writing it

1384
00:53:15,680 --> 00:53:17,040
myself.
What?

1385
00:53:17,040 --> 00:53:19,040
What would be your pushback to
someone who says that?

1386
00:53:19,280 --> 00:53:21,480
You're a teacher market.
Yeah, honestly, give it

1387
00:53:21,480 --> 00:53:22,800
feedback.
Just give it feedback.

1388
00:53:22,960 --> 00:53:25,280
Child B didn't really capture
this part of their their

1389
00:53:25,280 --> 00:53:27,160
personality.
Can you hone in on that and just

1390
00:53:27,160 --> 00:53:28,760
do a little bit less of this?
Okay, thanks.

1391
00:53:28,760 --> 00:53:30,400
I'll I'll regenerate child B for
you.

1392
00:53:30,400 --> 00:53:32,800
And but even in general, if it's
not doing what you want in

1393
00:53:32,800 --> 00:53:35,520
general, like to say they're too
long or it's not quite as

1394
00:53:35,520 --> 00:53:38,120
formal, just train it right.
Yeah.

1395
00:53:38,120 --> 00:53:39,240
Oh, yeah, yeah.
That's, that's what you're

1396
00:53:39,240 --> 00:53:39,920
getting at.
Absolutely.

1397
00:53:39,920 --> 00:53:40,640
Right.
You can put in.

1398
00:53:40,640 --> 00:53:42,400
I actually did this.
I probably should have mentioned

1399
00:53:42,400 --> 00:53:43,880
it.
I actually fed it a few examples

1400
00:53:43,880 --> 00:53:47,360
of my own personal handwritten,
handwritten but typed reports

1401
00:53:47,760 --> 00:53:50,000
and said this is my tone.
This is my style of writing.

1402
00:53:50,000 --> 00:53:51,560
Use it to help you.
And it did.

1403
00:53:51,600 --> 00:53:53,280
Yeah.
It was so good at doing that.

1404
00:53:53,280 --> 00:53:54,200
There's loads of things you can
do.

1405
00:53:54,200 --> 00:53:56,400
And at the end of the day, you
can always manually edit as

1406
00:53:56,400 --> 00:53:57,960
well.
Were there a few that I still

1407
00:53:57,960 --> 00:54:00,280
just took a sentence out and
replaced it because I was like,

1408
00:54:00,280 --> 00:54:03,000
I just, it be quicker for me to
just rewrite that sentence than

1409
00:54:03,000 --> 00:54:05,480
it will be to reprompt it?
Yeah, Just use your common sense

1410
00:54:05,480 --> 00:54:06,720
at that point.
And it was quick.

1411
00:54:06,720 --> 00:54:08,640
I did them all in under a couple
of hours.

1412
00:54:08,640 --> 00:54:12,640
And they were so precisely,
exactly what I wanted to say.

1413
00:54:12,640 --> 00:54:14,440
Yeah.
And I was almost like I.

1414
00:54:14,600 --> 00:54:17,080
I, I couldn't have written these
more personalised and better

1415
00:54:17,080 --> 00:54:18,960
myself.
This is just what I'm thinking,

1416
00:54:19,240 --> 00:54:21,800
but written perfectly well.
This is the issue I've always

1417
00:54:21,800 --> 00:54:24,840
found with writing reports is
that by far the biggest amount

1418
00:54:24,840 --> 00:54:28,240
of brain power goes on to
formulating how the paragraph

1419
00:54:28,240 --> 00:54:30,160
will be laid out and the
structure and making sure it

1420
00:54:30,160 --> 00:54:32,560
flows and it's written at a
certain formality, etcetera.

1421
00:54:32,720 --> 00:54:34,480
And doesn't just sound the same
every single time.

1422
00:54:34,560 --> 00:54:37,080
Because what ended up happening
before was Oh my God, by the

1423
00:54:37,080 --> 00:54:39,480
end, I'm I'm copying and pasting
changing names, right?

1424
00:54:39,480 --> 00:54:42,280
And that is the least personal
you can possibly go, whereas

1425
00:54:42,280 --> 00:54:44,200
this, the computer does it for
you.

1426
00:54:44,200 --> 00:54:45,880
It means immediately, and I want
to touch on what you said a

1427
00:54:45,880 --> 00:54:48,760
minute ago because I know for a
fact I've spoken about AI with

1428
00:54:48,760 --> 00:54:51,240
other teachers and they've said
it takes me just as long to do

1429
00:54:51,240 --> 00:54:52,960
those bullet points because
we've said you just use bullet

1430
00:54:52,960 --> 00:54:54,760
points.
Yeah, write down the headline

1431
00:54:54,760 --> 00:54:57,280
things that you think for each
child and the people going oh,

1432
00:54:57,280 --> 00:54:59,520
but that just takes me just as
long to think of and and maybe

1433
00:54:59,520 --> 00:55:01,680
I'll miss something, but also
like it takes me ages just to

1434
00:55:01,680 --> 00:55:03,800
write it.
There's a way around it.

1435
00:55:03,800 --> 00:55:07,560
You've spoken about something we
both do, which is before you do

1436
00:55:07,560 --> 00:55:10,240
anything, before you do anything
at all, it take any notes for

1437
00:55:10,240 --> 00:55:12,960
any children, write down, write
down.

1438
00:55:12,960 --> 00:55:15,280
Firstly, maybe a report of what
you want it to look like.

1439
00:55:15,360 --> 00:55:18,080
So spend, you know, 1520 minutes
writing it perfectly.

1440
00:55:18,240 --> 00:55:20,520
Yeah, because you can then
upload that later on to ChatGPT,

1441
00:55:20,520 --> 00:55:23,080
like you said.
But also alongside that, the way

1442
00:55:23,080 --> 00:55:24,960
I've always done reports, even
when I've been writing it

1443
00:55:24,960 --> 00:55:27,760
myself, is I've created a
structure to the report.

1444
00:55:27,760 --> 00:55:29,600
Yeah.
And it's an overarching theme

1445
00:55:29,960 --> 00:55:32,600
for each part.
For example, first bullet point

1446
00:55:33,080 --> 00:55:36,640
attitude in class, second bullet
point social circle and

1447
00:55:36,640 --> 00:55:39,640
friendship groups.
Third bullet point effort.

1448
00:55:39,720 --> 00:55:42,840
Yeah. 4th bullet point behaviour
in class and how they respond to

1449
00:55:42,840 --> 00:55:45,160
how they treat other adults and
other other people, for example.

1450
00:55:45,360 --> 00:55:48,800
5th bullet point, you know core
personal strength so.

1451
00:55:48,880 --> 00:55:50,400
You have that always there to
see.

1452
00:55:50,400 --> 00:55:51,680
That's a reference.
Exactly.

1453
00:55:51,680 --> 00:55:54,000
So when I've got when I've got
then child A, what I used to do

1454
00:55:54,000 --> 00:55:57,240
was go, I'm going to write a
sentence now about their, their

1455
00:55:57,240 --> 00:55:59,240
social bonds with the children.
Yeah.

1456
00:55:59,240 --> 00:56:01,400
And I'd sit there and I go, OK,
well, I know, I know I could

1457
00:56:01,400 --> 00:56:03,560
write a bullet point now.
I could write in 10 seconds.

1458
00:56:03,920 --> 00:56:08,680
A few strong individual friend
friendship bonds with children

1459
00:56:08,680 --> 00:56:11,640
in the class struggle sometimes
with making friends in larger

1460
00:56:11,640 --> 00:56:13,000
groups on playtime.
That's said there's.

1461
00:56:13,080 --> 00:56:14,320
A bullet point.
That's a bullet point.

1462
00:56:14,520 --> 00:56:18,040
It's not acceptable as a formal
written report because it's not

1463
00:56:18,040 --> 00:56:20,360
written in a formal way.
It's not written in a way, but

1464
00:56:20,360 --> 00:56:23,880
maybe that is sympathetic to the
child as much because I've just

1465
00:56:23,880 --> 00:56:26,280
written the fact that they're
not very good at making friends

1466
00:56:26,320 --> 00:56:28,040
in the playground.
I'd never put that in the actual

1467
00:56:28,040 --> 00:56:29,840
report.
That would take me 5 seconds.

1468
00:56:29,840 --> 00:56:32,240
So I know that information.
What would take me time

1469
00:56:32,240 --> 00:56:35,400
previously about AI would be
formulating that in a nicely

1470
00:56:35,400 --> 00:56:37,280
written way that isn't repeating
what I've said for every other

1471
00:56:37,280 --> 00:56:39,360
child in the same situation and
then moving on to the next

1472
00:56:39,360 --> 00:56:41,440
thing.
So the reason why it's really

1473
00:56:41,440 --> 00:56:43,960
powerful just to spend a bit of
time, I think is the thing here.

1474
00:56:44,160 --> 00:56:46,960
Spend a bit of time having a
plan, having your bullet points,

1475
00:56:47,200 --> 00:56:49,560
spend a bit of time just writing
it out and it'd be very quick

1476
00:56:49,560 --> 00:56:51,200
because you know your children
so well by the end of the year.

1477
00:56:51,200 --> 00:56:53,640
Through your bullet points.
And spend a bit of time just

1478
00:56:53,640 --> 00:56:57,280
training up the AI.
Don't make it create every

1479
00:56:57,280 --> 00:56:58,520
single report for you
immediately.

1480
00:56:58,520 --> 00:57:00,160
Don't do that.
Play around with one child.

1481
00:57:00,520 --> 00:57:02,720
Say what you're going to do the
prompts and say do it for this

1482
00:57:02,720 --> 00:57:04,560
child.
OK, Well, that OK, that doesn't

1483
00:57:04,560 --> 00:57:06,560
fit in the box that my school
will give me as a pro forma.

1484
00:57:06,720 --> 00:57:08,840
OK, do it again, but this time
140 words long.

1485
00:57:08,840 --> 00:57:10,600
Yeah, yeah, do it again, but
make it more formal.

1486
00:57:10,600 --> 00:57:12,760
Because even though I said in
formal tone, I know this is a

1487
00:57:12,760 --> 00:57:15,320
formal document, You played
around with it a bit.

1488
00:57:15,320 --> 00:57:17,320
Track GBT.
Instead of giving up, spend a

1489
00:57:17,320 --> 00:57:18,480
bit of time training.
Yeah.

1490
00:57:18,880 --> 00:57:21,960
And then when you're happy, then
you've written your notes for

1491
00:57:21,960 --> 00:57:23,920
the child.
It takes it takes honesty.

1492
00:57:24,480 --> 00:57:25,920
It takes me.
I'm going to be realistic

1493
00:57:25,920 --> 00:57:29,400
probably 1/5 of the time to
write bullet point notes about a

1494
00:57:29,400 --> 00:57:32,800
child compared to writing a
fully fledged out formal piece

1495
00:57:32,800 --> 00:57:36,160
of piece of writing a paragraph.
But it's still got 100% of the

1496
00:57:36,160 --> 00:57:37,120
personal touch.
Yeah.

1497
00:57:37,240 --> 00:57:40,320
Because all all that my brain
power is going on is formulating

1498
00:57:40,320 --> 00:57:41,240
it and writing it in a certain
way.

1499
00:57:41,240 --> 00:57:42,320
Yeah, Yeah, yeah.
That's where the large.

1500
00:57:42,360 --> 00:57:44,240
Language model comes in just
cutting out the last step where

1501
00:57:44,240 --> 00:57:46,320
you would have turned your
informal bullet point in your

1502
00:57:46,320 --> 00:57:47,960
head into your own formal
sentence.

1503
00:57:47,960 --> 00:57:49,560
You're now just saying, here's
my informal bullet point.

1504
00:57:49,560 --> 00:57:52,080
Can you just do the last bit for
me and turn it into a formal

1505
00:57:52,080 --> 00:57:53,120
sentence?
Because that's a bit that

1506
00:57:53,120 --> 00:57:55,880
doesn't really matter about us.
Who cares if we do that?

1507
00:57:55,880 --> 00:57:56,920
But it just needs to be
personal.

1508
00:57:57,160 --> 00:57:59,200
That's very different to the
opinion a couple of years ago.

1509
00:57:59,200 --> 00:58:00,760
People like we can't use AIT
reports.

1510
00:58:00,760 --> 00:58:04,200
They had this idea that we were
literally saying here is here's

1511
00:58:04,200 --> 00:58:07,160
32 names, right, missing reports
and it's just completely

1512
00:58:07,160 --> 00:58:08,600
made-up.
It's like, do you know what?

1513
00:58:08,600 --> 00:58:10,520
I, I can't imagine anyone was
even doing that.

1514
00:58:10,520 --> 00:58:13,000
Then I think people just made-up
some some circumstances like no

1515
00:58:13,000 --> 00:58:14,040
one's doing that what you're
talking.

1516
00:58:14,040 --> 00:58:16,200
About and and if they did you
know what you would do to that

1517
00:58:16,200 --> 00:58:18,720
person you would talk to that
person and explain why that's

1518
00:58:18,720 --> 00:58:20,920
bad and get them to do it in a
different way What you wouldn't

1519
00:58:20,920 --> 00:58:22,920
do was just take away this
mechanism from everyone in the

1520
00:58:22,920 --> 00:58:26,200
whole country yeah yeah classic,
classic, classic like that it's

1521
00:58:26,200 --> 00:58:29,120
crazy yeah.
I think understanding ChatGPT in

1522
00:58:29,120 --> 00:58:32,120
general is going to be able to
unlock I think a lot of these

1523
00:58:32,120 --> 00:58:33,480
for.
Can I add one thing to it as

1524
00:58:33,480 --> 00:58:35,160
well, because it's important to
understand this again, I'm

1525
00:58:35,160 --> 00:58:36,840
really I'm really here to try
and convert people that maybe

1526
00:58:36,840 --> 00:58:39,080
had a dabble in this last year
and weren't completely

1527
00:58:39,080 --> 00:58:41,480
convinced, right.
ChatGPT since then has developed

1528
00:58:41,480 --> 00:58:43,400
its memory system really,
really, really well.

1529
00:58:43,400 --> 00:58:46,000
So ChatGPT 3 was the first one
that came out in 2022.

1530
00:58:46,040 --> 00:58:48,040
It actually didn't have memory.
It didn't have a memory system,

1531
00:58:48,120 --> 00:58:50,640
right.
Whereas now if you if you have

1532
00:58:50,640 --> 00:58:53,600
like 1 long stream of chat with
ChatGPT or even another whole

1533
00:58:53,600 --> 00:58:56,160
new chat, it kind of starts
learning you as a person, right.

1534
00:58:56,440 --> 00:58:59,280
So if you've done all this grunt
work, like Dylan said with one

1535
00:58:59,280 --> 00:59:01,760
just to get the report perfectly
how you said it is, you don't

1536
00:59:01,760 --> 00:59:04,000
then have to re prompt it
underneath and be like, OK, I

1537
00:59:04,520 --> 00:59:06,680
want you to do this, this, this,
this, this and this and this for

1538
00:59:06,680 --> 00:59:09,960
all my other kids.
You just say do that for this

1539
00:59:09,960 --> 00:59:11,520
whole list now.
And it knows what that is

1540
00:59:11,520 --> 00:59:13,880
because it has it's it's
understanding of reasoning and

1541
00:59:13,880 --> 00:59:16,640
it's understanding of even
nuance is is crazy good.

1542
00:59:16,640 --> 00:59:18,880
Now, even like one more thing,
just I'm trying to think of

1543
00:59:18,880 --> 00:59:20,360
anything that people might have
against this.

1544
00:59:20,360 --> 00:59:22,280
Right.
When you said your bullet point

1545
00:59:22,280 --> 00:59:24,480
might be quite harsh and to the
point and you obviously wouldn't

1546
00:59:24,480 --> 00:59:26,120
word it like that.
People might say, well, if

1547
00:59:26,120 --> 00:59:28,680
you're just giving ChatGPT A
harsh point, it might word it

1548
00:59:28,680 --> 00:59:31,520
with your harsh wordings.
Nope, It is so good at

1549
00:59:31,520 --> 00:59:33,320
understanding nuance.
Now it knows that's rude.

1550
00:59:33,840 --> 00:59:35,920
I know it sounds crazy.
Me saying it's just, it's just

1551
00:59:35,920 --> 00:59:38,200
an AI bot.
It knows that's rude and it will

1552
00:59:38,200 --> 00:59:40,360
not word it like.
That and if it doesn't retrain

1553
00:59:40,360 --> 00:59:42,680
it, that's the point.
If it doesn't do that, just say

1554
00:59:42,680 --> 00:59:45,960
to it that was to on the nose.
Although we still want this

1555
00:59:45,960 --> 00:59:48,560
truth to come through because it
is something that a parent

1556
00:59:48,560 --> 00:59:50,600
should know about a child if
they're struggling socially.

1557
00:59:51,040 --> 00:59:52,960
Can we do it in a way that's
more sensitive, perhaps, but

1558
00:59:52,960 --> 00:59:55,480
still formal, And it will do it.
Yeah, and the chances are if you

1559
00:59:55,480 --> 00:59:56,920
get your prompt right at the
start, you won't have to.

1560
00:59:56,920 --> 00:59:59,480
I think in my prompt I put
remember this is has to be

1561
00:59:59,480 --> 01:00:01,160
appropriate for parents reading
about their children.

1562
01:00:01,160 --> 01:00:02,760
Yeah, something like that along
the lines.

1563
01:00:02,760 --> 01:00:05,840
So and it just, it just
extrapolated that OK, this needs

1564
01:00:05,840 --> 01:00:08,160
to be appropriate because it's
it knows everything about every

1565
01:00:08,160 --> 01:00:10,640
text in the Internet ever.
It knows what it looks like.

1566
01:00:10,640 --> 01:00:13,320
It's just, it's insanely good.
I I couldn't recommend it

1567
01:00:13,320 --> 01:00:13,840
anymore.
I.

1568
01:00:13,840 --> 01:00:15,840
Think that's that That is the
big point though, isn't it about

1569
01:00:15,920 --> 01:00:18,520
spending time getting it to the
level it should be?

1570
01:00:18,640 --> 01:00:20,360
Yeah.
But of course, Dylan, there are

1571
01:00:20,360 --> 01:00:23,400
always drawbacks, right?
There are, There are, there are

1572
01:00:23,400 --> 01:00:25,960
things that people will
counteract this with that there

1573
01:00:25,960 --> 01:00:27,160
are arguments.
People say, yeah, but we

1574
01:00:27,160 --> 01:00:28,720
shouldn't be using AI for this
reason.

1575
01:00:28,800 --> 01:00:29,800
Yeah.
Should we go through a few of

1576
01:00:29,800 --> 01:00:30,360
those?
And.

1577
01:00:30,360 --> 01:00:32,480
And yeah, I feel like we've
covered quite a few of them.

1578
01:00:32,480 --> 01:00:35,000
Yeah, because we talked about
GDPRGDPR is a big problem.

1579
01:00:35,000 --> 01:00:37,000
We don't want to be uploading
names that's identifiable for

1580
01:00:37,000 --> 01:00:39,360
children.
Easy way around that is simply

1581
01:00:39,600 --> 01:00:41,280
if you, especially if you're
going down the free route of

1582
01:00:41,280 --> 01:00:44,600
using ChatGPT yourself, just
build in that yourself, label

1583
01:00:44,600 --> 01:00:46,880
the children like you said,
ABCD, go back in and just put a

1584
01:00:46,880 --> 01:00:48,120
name in.
It's still quicker.

1585
01:00:48,360 --> 01:00:51,840
I think you can avoid GDPR risks
and by taking away any

1586
01:00:51,840 --> 01:00:53,720
identifiable names with the
children.

1587
01:00:53,960 --> 01:00:55,360
I think that's then fine.
Yeah.

1588
01:00:55,360 --> 01:00:57,120
The fact it takes so long, we've
talked about that.

1589
01:00:57,120 --> 01:01:00,040
I just don't think that's true.
If you're doing it properly, if

1590
01:01:00,040 --> 01:01:02,640
you're doing it in a way which
saves time, obviously it's gonna

1591
01:01:02,640 --> 01:01:03,480
save time.
Yeah.

1592
01:01:03,480 --> 01:01:06,480
If you're essentially writing it
and then just putting it into

1593
01:01:06,480 --> 01:01:08,600
chat TPT and it just changes
what you've already written as a

1594
01:01:08,600 --> 01:01:10,960
fully fledged paragraph, yeah,
that won't save you time.

1595
01:01:10,960 --> 01:01:13,560
You're adding an extra step in.
So just be sensible about what

1596
01:01:13,560 --> 01:01:15,080
you're doing.
I think there's a couple of

1597
01:01:15,080 --> 01:01:18,480
things we can touch upon here
that we haven't maybe, but we

1598
01:01:18,480 --> 01:01:20,720
can briefly talk about now
because there'll be people who

1599
01:01:20,720 --> 01:01:23,280
sat through this whole thing
going AI is, yeah, it's

1600
01:01:23,280 --> 01:01:24,560
brilliant.
It is fantastic.

1601
01:01:24,560 --> 01:01:26,840
It does all this stuff.
It can have like beautiful

1602
01:01:26,840 --> 01:01:29,840
outcomes for the children.
I spoke earlier about how that

1603
01:01:29,840 --> 01:01:31,720
child's now engaged to do
writing.

1604
01:01:31,720 --> 01:01:35,640
Isn't that amazing?
But isn't it's awful for the

1605
01:01:35,640 --> 01:01:36,880
environment?
Yeah, There are just things

1606
01:01:36,880 --> 01:01:38,560
where it's like, I don't have an
answer for this, but we should.

1607
01:01:38,560 --> 01:01:41,160
We should acknowledge them
because if we're here, we should

1608
01:01:41,160 --> 01:01:43,160
be responsible.
Like, yeah, it is it.

1609
01:01:43,640 --> 01:01:45,360
Using energy that we otherwise
wouldn't be.

1610
01:01:45,360 --> 01:01:47,000
Using it uses a lot of energy, a
lot more.

1611
01:01:47,120 --> 01:01:49,760
I think people realise that to
generate one prompt, it's a lot

1612
01:01:49,760 --> 01:01:52,440
of computing power and computing
power generally generates heat

1613
01:01:52,680 --> 01:01:55,280
and to how do we, how do we
generally cool down computing

1614
01:01:55,280 --> 01:01:58,400
systems Normally with water, we
and we need a lot of water to do

1615
01:01:58,400 --> 01:01:59,360
it.
And do you know what I mean?

1616
01:01:59,360 --> 01:02:01,120
The whole system goes there.
It does create a lot of energy.

1617
01:02:01,120 --> 01:02:03,400
So it's worth considering.
I think a lot of AI companies

1618
01:02:03,400 --> 01:02:07,720
are are trying to build their
new infrastructure with this in

1619
01:02:07,720 --> 01:02:11,040
mind to like reduce it as much.
But at the end of the day, it

1620
01:02:11,040 --> 01:02:12,960
will always use loads more
energy.

1621
01:02:12,960 --> 01:02:14,120
They're not.
They're not ever going to get to

1622
01:02:14,120 --> 01:02:15,600
the point where they're like,
oh, it doesn't use energy

1623
01:02:15,600 --> 01:02:17,640
anymore, Of course it will.
They might create better cooling

1624
01:02:17,640 --> 01:02:20,400
systems.
I'm not suddenly not sceptical

1625
01:02:20,400 --> 01:02:21,920
about big business, just I like
AI.

1626
01:02:21,920 --> 01:02:24,080
Yeah, yeah, I'm not going to be
like they'll give me a free pass

1627
01:02:24,080 --> 01:02:25,400
because it does this really cool
thing for me.

1628
01:02:25,400 --> 01:02:27,960
No, no, this is huge business
making billions of pounds.

1629
01:02:28,120 --> 01:02:29,880
They're going to do shortcuts
and they're going to rinse

1630
01:02:29,880 --> 01:02:30,720
energy.
And that is a.

1631
01:02:30,720 --> 01:02:32,000
Problem.
It's a huge challenge and it's a

1632
01:02:32,000 --> 01:02:33,440
huge downfall and a downside to
use.

1633
01:02:33,440 --> 01:02:36,760
Is it enough to not use?
It depends where you stand on

1634
01:02:36,760 --> 01:02:38,480
it, I suppose, isn't it?
For some people it is do.

1635
01:02:38,480 --> 01:02:42,920
You drive to the shop, you know
driving is is 100 times more

1636
01:02:42,920 --> 01:02:45,120
than doing prompts.
Yeah, I think I read somewhere

1637
01:02:45,800 --> 01:02:50,000
that per kilometre driving in a
petrol car is equivalent to

1638
01:02:50,000 --> 01:02:54,680
about 250 AI prompts, right.
OK, so I'm not, I'm not

1639
01:02:54,680 --> 01:02:57,600
justifying stuff here because
because again, using AI, using

1640
01:02:57,640 --> 01:03:00,640
AI is like I think 15 to 20
times more energy than using

1641
01:03:00,640 --> 01:03:02,280
Google.
So there's obviously layers

1642
01:03:02,280 --> 01:03:03,480
here.
I'm don't get me wrong.

1643
01:03:03,520 --> 01:03:05,400
And the small things are still
important.

1644
01:03:05,440 --> 01:03:07,080
They are.
I'm not saying they aren't.

1645
01:03:07,360 --> 01:03:12,000
But if you can, for example, if
you're working from home, you've

1646
01:03:12,000 --> 01:03:15,440
chosen to work from home and
you're not no longer driving 10

1647
01:03:15,440 --> 01:03:20,000
kilometres there and back, OK?
That is, you've got over 2000 AI

1648
01:03:20,000 --> 01:03:22,920
prompts you can use that day and
still be in the green for how

1649
01:03:22,920 --> 01:03:25,360
much energy you've consumed.
Yeah, because you've saved the

1650
01:03:25,360 --> 01:03:28,960
driving energy for doing that.
It's just a game of give and

1651
01:03:28,960 --> 01:03:31,600
take and I think you've just got
to be sensible about it,

1652
01:03:31,760 --> 01:03:35,280
understand what you're doing, be
be able to defend it in the same

1653
01:03:35,280 --> 01:03:37,320
way that I defend driving to
Tesco, which is 2 minutes down

1654
01:03:37,320 --> 01:03:38,560
the road.
I'll do that.

1655
01:03:38,560 --> 01:03:40,520
Sometimes I'll walk, but
sometimes I'm like, no, it'll

1656
01:03:40,520 --> 01:03:42,080
take half the time.
I'll just drive quickly.

1657
01:03:42,520 --> 01:03:45,520
I think it's easy to forget that
that like that compared to

1658
01:03:45,520 --> 01:03:47,200
something new.
When something new comes in,

1659
01:03:47,200 --> 01:03:49,560
it's like, Oh no, no, it's using
energy that wasn't there before.

1660
01:03:49,760 --> 01:03:53,080
Yeah, But there are other things
I can do in my life if I'm

1661
01:03:53,080 --> 01:03:56,320
really, truly worried about that
level of energy usage.

1662
01:03:56,800 --> 01:03:59,800
That can easily compensate for
that Yeah, yeah, just to do 1

1663
01:03:59,800 --> 01:04:02,280
less drive a month, honestly,
and that'll probably account for

1664
01:04:02,280 --> 01:04:04,280
all the amount of times you use
ChatGPT as a normal, but as a

1665
01:04:04,280 --> 01:04:05,560
normal user.
That's fair enough.

1666
01:04:05,560 --> 01:04:07,600
And the and the last big one, I
think we kind of covered it, but

1667
01:04:07,600 --> 01:04:09,160
it's worth mentioning.
It's just that people say it's

1668
01:04:09,160 --> 01:04:12,120
just morally wrong.
It's morally wrong to not write

1669
01:04:12,120 --> 01:04:13,680
it yourself.
These are reports that are going

1670
01:04:13,680 --> 01:04:15,880
to parents.
It shouldn't be under the

1671
01:04:15,880 --> 01:04:19,080
disguise that you didn't write
it entirely yourself.

1672
01:04:19,080 --> 01:04:20,400
It's a big argument that people
make.

1673
01:04:21,160 --> 01:04:22,640
I I don't think it has many
legs.

1674
01:04:22,640 --> 01:04:25,200
What do you think about that?
Morally wrong to use AI?

1675
01:04:25,760 --> 01:04:28,560
Did I hope this person turned
off spell check when they used

1676
01:04:28,560 --> 01:04:30,160
their Word document to write
their reports?

1677
01:04:30,400 --> 01:04:33,200
I hope this person never looked
up a synonym for a word like

1678
01:04:33,200 --> 01:04:35,640
what?
At what point does everything

1679
01:04:35,640 --> 01:04:36,920
have to come from your own
brain?

1680
01:04:37,480 --> 01:04:40,960
I think what has to come from
your own brain is the fact and

1681
01:04:40,960 --> 01:04:44,320
information about the children
that has to come from the person

1682
01:04:44,320 --> 01:04:45,360
who's been with the child all
year.

1683
01:04:45,760 --> 01:04:49,560
That's just a fact.
The the prose and the way we

1684
01:04:49,560 --> 01:04:53,360
write it and the formality of it
and the word choice of it

1685
01:04:53,880 --> 01:04:56,080
absolutely doesn't have to come
from a human.

1686
01:04:56,080 --> 01:04:58,160
And it never has done.
No, it's never been.

1687
01:04:58,440 --> 01:05:01,080
We used to have banks.
I think Twinkle used to release

1688
01:05:01,080 --> 01:05:03,880
a bank or they still do, where
you would take sentence stems

1689
01:05:04,080 --> 01:05:06,520
because you, the adult, have
chosen which sentence stem

1690
01:05:06,520 --> 01:05:08,080
matches the child.
That's your input.

1691
01:05:08,200 --> 01:05:10,160
But it doesn't matter how it's
written and if the same child

1692
01:05:10,160 --> 01:05:11,360
gets the same sentence.
Yeah, they just need the

1693
01:05:11,360 --> 01:05:12,640
information.
They just need the information.

1694
01:05:12,640 --> 01:05:14,960
So I think there's a clear
distinction, and that always has

1695
01:05:15,000 --> 01:05:17,560
been between what clearly has to
come from the teacher and what

1696
01:05:17,560 --> 01:05:19,240
doesn't.
Yeah, we're just adding AI to

1697
01:05:19,240 --> 01:05:21,480
this side of things that it
doesn't have to, making it

1698
01:05:21,480 --> 01:05:24,200
cleaner, making it way more
personal, making it better than

1699
01:05:24,200 --> 01:05:26,600
what already existed.
But we're kind of just some

1700
01:05:26,600 --> 01:05:28,280
people are seeing AI and going
morally wrong.

1701
01:05:28,280 --> 01:05:31,360
I was like, well, it's no less
wrong than using a Bank of

1702
01:05:31,360 --> 01:05:33,040
words.
Yeah, you're still choosing as

1703
01:05:33,040 --> 01:05:34,240
the adult.
And if anything, it's better

1704
01:05:34,240 --> 01:05:36,560
because it's more personal.
It's funny what the perception

1705
01:05:36,560 --> 01:05:38,000
of being professional does,
doesn't it?

1706
01:05:38,080 --> 01:05:40,760
Because really, if we go down
this Fort train, why aren't we

1707
01:05:40,760 --> 01:05:41,880
just giving them the bullet
points?

1708
01:05:42,120 --> 01:05:44,440
Yeah.
Why aren't we just giving the

1709
01:05:44,440 --> 01:05:46,040
parents the bullet points that
we write that we put into

1710
01:05:46,040 --> 01:05:48,040
ChatGPT and they have to
decipher it back into bullet

1711
01:05:48,040 --> 01:05:49,160
points?
Because that's basically what

1712
01:05:49,160 --> 01:05:50,800
they're doing is they're reading
it, they're interpreting it and

1713
01:05:50,800 --> 01:05:52,800
then going, oh, OK, so my kids.
Are so the bullet points are

1714
01:05:52,800 --> 01:05:54,520
this, yeah.
They go my kids a bit silly in

1715
01:05:54,520 --> 01:05:56,080
class, but obviously the
professionals, they've worded it

1716
01:05:56,080 --> 01:05:57,040
nicely.
Yeah, it's funny enough, mate.

1717
01:05:57,040 --> 01:05:58,480
I actually put that in my
original bullet point.

1718
01:05:58,480 --> 01:05:59,440
Your kid is a bit silly in
class.

1719
01:06:00,200 --> 01:06:02,200
Yeah, Doesn't this, I think
there's a meta conversation

1720
01:06:02,200 --> 01:06:04,000
maybe just to finish this off
when I was like 30 seconds.

1721
01:06:04,000 --> 01:06:05,960
Yeah.
How useful are reports anyway?

1722
01:06:05,960 --> 01:06:08,120
We've spoken about this.
What are reports?

1723
01:06:08,240 --> 01:06:11,040
To me, they're like a keepsake
to see what you were like when

1724
01:06:11,040 --> 01:06:12,120
you were younger, when you're
older.

1725
01:06:12,360 --> 01:06:15,720
And and they're statutory needed
to give to the parents of

1726
01:06:15,840 --> 01:06:17,160
government.
Say you have to report on the

1727
01:06:17,160 --> 01:06:19,200
children.
They don't have to be all

1728
01:06:19,200 --> 01:06:20,960
singing and dancing.
No, from school to school

1729
01:06:20,960 --> 01:06:22,480
reports look different.
I've worked in a school where

1730
01:06:22,480 --> 01:06:25,880
you got like a 30 page booklet
for your report and there was

1731
01:06:25,880 --> 01:06:29,080
some massively in depth stuff.
And we as teachers, you had to

1732
01:06:29,080 --> 01:06:30,760
write so much.
And that was before AI.

1733
01:06:31,080 --> 01:06:32,800
And I've worked in another
school where it's one side of a

1734
01:06:32,840 --> 01:06:35,360
four.
So, so like there's, there's a

1735
01:06:35,360 --> 01:06:37,520
discrepancy there.
Wait, it's coming full circle

1736
01:06:37,520 --> 01:06:38,880
here.
Because it seems to me like,

1737
01:06:38,880 --> 01:06:41,720
yes, there are sometimes things
in schools that are mental that

1738
01:06:41,720 --> 01:06:44,640
we simply don't have to do, but
we convince ourselves we do for

1739
01:06:44,640 --> 01:06:46,760
some reason.
And then that from the knock on

1740
01:06:46,760 --> 01:06:49,080
effect of someone in charge
who's decided that means

1741
01:06:49,160 --> 01:06:51,600
hundreds and hundreds of hours
for teachers to have to do in

1742
01:06:51,600 --> 01:06:52,840
their own time.
Guess what?

1743
01:06:52,840 --> 01:06:55,880
Guys use AI.
My school said I can't use AI.

1744
01:06:56,080 --> 01:06:57,720
Do it anyway.
They're not going to know.

1745
01:06:57,720 --> 01:06:59,240
They don't come full circle on
this episode.

1746
01:06:59,240 --> 01:07:01,520
You just wait, give it a couple
of years, and moderations won't

1747
01:07:01,520 --> 01:07:02,640
exist.
You know what's going to happen?

1748
01:07:02,840 --> 01:07:05,080
Everyone has to upload their
children's writing into this

1749
01:07:05,080 --> 01:07:07,520
insane AI machine that knows the
teacher assessment framework and

1750
01:07:07,520 --> 01:07:08,920
it tells you whether it's
expected or not.

1751
01:07:08,920 --> 01:07:11,160
CLA everybody.
No one's got a job anymore and

1752
01:07:11,160 --> 01:07:13,000
this episode is over.
Yeah, thank you, everyone.

1753
01:07:13,000 --> 01:07:17,800
Potential, literally, literally.
And yeah, I just want to spread

1754
01:07:17,800 --> 01:07:20,400
my thanks everyone for watching
this AI generated piece of

1755
01:07:20,400 --> 01:07:23,000
content.
If you enjoyed this AI generated

1756
01:07:23,000 --> 01:07:24,960
podcast, please leave us a
review.

1757
01:07:25,120 --> 01:07:26,680
See you next time.
Bye.

1758
01:07:26,720 --> 01:07:27,800
I'm real, I promise.