1
00:00:06,200 --> 00:00:10,240
Welcome to Thinking Deeply About
Primary Education, the podcast 

2
00:00:10,240 --> 00:00:14,040
that makes time and space to 
think about pedagogy, teaching 

3
00:00:14,040 --> 00:00:15,800
and learning professional 
development. 

4
00:00:16,040 --> 00:00:18,520
Anything of interest to time. 
Poor but enthusiastic primary 

5
00:00:18,520 --> 00:00:21,920
teachers join us as we delve 
into the murky depths of 

6
00:00:21,920 --> 00:00:24,880
education in an attempt to help 
you enhance your classroom 

7
00:00:24,880 --> 00:00:45,490
practice. 
In today's episode, I, Neil 

8
00:00:45,490 --> 00:00:51,850
Almond will be taking the reins.
Welcome to Thinking DV by 

9
00:00:51,850 --> 00:00:54,130
Primary Education, the podcast 
that makes time and space to 

10
00:00:54,130 --> 00:00:56,890
think about pedagogy, teaching 
and learning, professional 

11
00:00:56,890 --> 00:00:58,730
development. 
Anything of interest to Time, 

12
00:00:58,730 --> 00:01:00,810
Pure but enthusiasm, rich 
primary teachers. 

13
00:01:01,130 --> 00:01:02,690
This week I'm joined by Neil 
Almond. 

14
00:01:03,450 --> 00:01:06,930
Needs to be back, Kieran. 
This week we're going to explore

15
00:01:07,410 --> 00:01:12,210
the potential of artificial 
intelligence and the impact it 

16
00:01:12,210 --> 00:01:18,250
might have on the work of 
teachers. 

17
00:01:18,490 --> 00:01:20,610
But first, Dale, what you 
reading for? 

18
00:01:23,210 --> 00:01:30,450
Hey, what you reading for? 
So I was recommended a book 

19
00:01:31,290 --> 00:01:36,810
called Spelling, Structure and 
Strategies by Paul and Jean 

20
00:01:36,810 --> 00:01:42,450
Hannah and Richard Hodges. 
It's quite an old book, 1971 

21
00:01:42,450 --> 00:01:45,330
quite so. 
It's getting on just over 50 

22
00:01:45,330 --> 00:01:51,920
years old now, quite enjoyable 
in the sense that it's affirming

23
00:01:51,920 --> 00:01:55,040
many of the biases that I have 
into how I think we should teach

24
00:01:55,040 --> 00:01:59,320
spelling, which on one hand is 
really good and on another hand,

25
00:01:59,320 --> 00:02:02,680
you know, not so great. 
What was interesting about this 

26
00:02:02,680 --> 00:02:05,680
one though is that they were 
kind of the first people there 

27
00:02:05,680 --> 00:02:06,960
was a study and I had no idea 
about this. 

28
00:02:06,960 --> 00:02:10,479
They did a study in the 60s and 
you know used the all the 

29
00:02:10,479 --> 00:02:14,880
computing power that you know we
had back in the 60s which 

30
00:02:14,880 --> 00:02:16,520
obviously was enough to put a 
man on the moon. 

31
00:02:16,520 --> 00:02:21,160
So fair amount of it to really 
kind of study the structure of 

32
00:02:21,160 --> 00:02:24,520
the English language. 
And so they're kind of the main 

33
00:02:24,520 --> 00:02:28,360
appeal for this book for me was 
around that structure element 

34
00:02:28,360 --> 00:02:34,120
because they're kind of in a 
very basic way, it would have to

35
00:02:34,120 --> 00:02:40,160
be updated for the modern era, 
kind of revealed the statistical

36
00:02:40,160 --> 00:02:42,160
properties of the English 
language. 

37
00:02:42,160 --> 00:02:48,480
So you could teach children to a
high degree percentage that if 

38
00:02:48,480 --> 00:02:55,840
this sound, the sound L was in 
the end position, then there was

39
00:02:56,040 --> 00:03:00,240
a I'm making this statistic up. 
So like a 95% chance that it's 

40
00:03:00,240 --> 00:03:03,920
spelt OW and that's kind of 
something that has always kind 

41
00:03:03,960 --> 00:03:06,080
of interested me and it's 
something that I felt has been 

42
00:03:06,080 --> 00:03:11,600
missing from spelling 
instruction that kind of making 

43
00:03:11,600 --> 00:03:17,160
that statistical elements 
explicit to students as quickly 

44
00:03:17,160 --> 00:03:19,000
as possible. 
So yeah, haven't finished it 

45
00:03:19,000 --> 00:03:21,480
yet, but really enjoyed kind of 
going through it and kind of 

46
00:03:21,480 --> 00:03:25,440
gives me hope that there'll be 
some researchers out there who 

47
00:03:25,440 --> 00:03:30,480
will create a larger corpus than
the 17,000 words I think that 

48
00:03:30,480 --> 00:03:35,000
they used when they did this 
ideally, hopefully British 

49
00:03:36,200 --> 00:03:38,400
English spelling as well. 
And obviously with more 

50
00:03:38,400 --> 00:03:42,440
computing power, stronger 
algorithms, you know, we'll be 

51
00:03:42,440 --> 00:03:46,400
able to produce something more 
up to date. 

52
00:03:46,400 --> 00:03:48,920
So we'll see what about you? 
What are you reading for? 

53
00:03:49,120 --> 00:03:51,040
Nice. 
Just the 50 years have passed. 

54
00:03:51,040 --> 00:03:53,880
Yeah. 
So I mean might be implemented 

55
00:03:53,880 --> 00:04:00,480
in classrooms any day now. 
I've got a maths one for 

56
00:04:00,560 --> 00:04:02,320
balance. 
It was shared with me. 

57
00:04:02,320 --> 00:04:04,760
You think? 
Oh, Brendan Lee shared it. 

58
00:04:05,440 --> 00:04:08,360
It's called The Need for Speed. 
Why Fluency Counts for Maths. 

59
00:04:08,360 --> 00:04:11,440
Learning by Tony Hat and Rob 
very interesting read. 

60
00:04:12,120 --> 00:04:14,080
I won't give away too many 
spoilers but I'm sure you can 

61
00:04:14,080 --> 00:04:17,839
probably guess what the sort of 
the the main takeaways are. 

62
00:04:17,880 --> 00:04:19,959
But yeah, well, well, we're 
checking out. 

63
00:04:23,040 --> 00:04:27,120
So I think probably Neil, it was
around about this time last year

64
00:04:27,320 --> 00:04:33,440
that you introduced me to Open 
AI and. 

65
00:04:33,920 --> 00:04:37,520
In that in those 12 months, a 
lot has happened. 

66
00:04:37,920 --> 00:04:41,880
The rest of everyone has tried 
to catch up with you, so 

67
00:04:42,120 --> 00:04:44,160
obviously you're that you're 
trail business in terms of 

68
00:04:44,160 --> 00:04:46,640
artificial intelligence. 
So I thought we'd sit down and 

69
00:04:46,640 --> 00:04:51,680
we'd look at what this might 
mean for teachers, because there

70
00:04:51,680 --> 00:04:54,760
are plenty of snack oil 
merchants who are looking to 

71
00:04:54,760 --> 00:04:58,080
make a fast buck off the back of
this. 

72
00:04:58,080 --> 00:05:00,160
And I promise I'm not one of 
them. 

73
00:05:00,720 --> 00:05:02,760
Well, now you you have no skin 
in the game. 

74
00:05:03,520 --> 00:05:09,400
And you know and you know the 
the the the charlatans aren't 

75
00:05:09,400 --> 00:05:10,800
there. 
They're not teachers. 

76
00:05:11,000 --> 00:05:14,040
And so I'm not commenting on any
educators here in in this 

77
00:05:14,040 --> 00:05:16,600
operating in this field, but 
rather you can see the promoted 

78
00:05:16,600 --> 00:05:18,960
tweets on Twitter and stuff like
that. 

79
00:05:18,960 --> 00:05:22,040
There are words like, you know, 
you know, I'm sure people have 

80
00:05:22,040 --> 00:05:25,960
seen them and, you know, so like
reskin and chat, GTV trying. 

81
00:05:25,960 --> 00:05:27,520
I don't know. 
Am I am I offending any 

82
00:05:27,560 --> 00:05:29,720
teachers? 
No, no, you're fine. 

83
00:05:29,720 --> 00:05:37,160
I'd say it's 80% of my Twitter 
feed now is just these promoted 

84
00:05:38,000 --> 00:05:41,120
accounts that are telling me 
that I could earn thousands of 

85
00:05:41,120 --> 00:05:44,800
dollars in side hustles if I 
just learn how to do these five 

86
00:05:44,800 --> 00:05:49,520
things in ChatGPT or Dolly or 
Mid Journey. 

87
00:05:49,520 --> 00:05:54,480
And yeah, it's it's a pain for 
my Twitter feed, Let's put it 

88
00:05:54,480 --> 00:05:57,120
that way. 
So I thought we'd explore it 

89
00:05:57,120 --> 00:05:59,040
because there must be some 
value, because there's a whole 

90
00:05:59,040 --> 00:06:02,000
lot of money being invested in 
this and around the world. 

91
00:06:03,240 --> 00:06:05,480
But it makes sense to start with
definitions of what we're 

92
00:06:05,480 --> 00:06:08,640
talking about. 
What is artificial intelligence?

93
00:06:10,080 --> 00:06:14,080
OK, so AI Kind of. 
For me it's this. 

94
00:06:14,080 --> 00:06:20,120
The study of using data in 
ridiculously large quantities to

95
00:06:21,080 --> 00:06:28,280
attempt to program intelligent 
machines that mimic human like 

96
00:06:28,480 --> 00:06:31,760
behaviour. 
That's my the definition that 

97
00:06:31,760 --> 00:06:37,240
I'm kind of going for here. 
Obviously the purpose of it I 

98
00:06:37,240 --> 00:06:42,240
guess is to improve how quickly,
accurately and successfully it 

99
00:06:42,240 --> 00:06:47,320
can complete tasks that we may 
ask, you know, humans to do 

100
00:06:47,680 --> 00:06:50,960
right now. 
We're kind of seeing the 

101
00:06:50,960 --> 00:06:57,520
beginnings of that sort of 
journey and attempt to mixed 

102
00:06:57,920 --> 00:07:00,640
success, I think is is fair to 
say. 

103
00:07:00,960 --> 00:07:03,480
I mean, when I think of 
artificial intelligence, I think

104
00:07:03,480 --> 00:07:07,680
of Was it Deep Blue? 
The chess? 

105
00:07:07,680 --> 00:07:08,960
Playing a robot? 
Yeah. 

106
00:07:10,120 --> 00:07:13,200
That that taught itself 
essentially through running 

107
00:07:13,200 --> 00:07:15,440
through millions and millions of
different games. 

108
00:07:16,240 --> 00:07:18,920
Hard to play chess to the point 
where it was unbeatable. 

109
00:07:18,920 --> 00:07:20,960
Was that was it deep blue? 
I mean, I know they they put 

110
00:07:20,960 --> 00:07:23,480
deep in front of, put deep into 
everything. 

111
00:07:23,760 --> 00:07:28,800
The reason it's called Deep Blue
is that particular sort of so AI

112
00:07:28,800 --> 00:07:31,440
is like a bit of like a good 
like umbrella term for lots of 

113
00:07:31,440 --> 00:07:38,160
other kind of other techniques 
that are AI, but there is 

114
00:07:38,160 --> 00:07:40,160
definitely a sort of like 
hierarchy to them. 

115
00:07:40,160 --> 00:07:45,040
So AI is sort of the the holding
concept, the overall sort of 

116
00:07:45,040 --> 00:07:48,760
umbrella and then you kind of 
have things like you know 

117
00:07:48,760 --> 00:07:52,160
machine learning which is a 
method of making predictions 

118
00:07:53,200 --> 00:07:57,960
with that doesn't sort of use 
predetermined human coded rules 

119
00:07:58,400 --> 00:08:01,440
and that kind of analysis, vast 
volumes of datas to find 

120
00:08:01,440 --> 00:08:03,280
patterns then kind of construct 
a model. 

121
00:08:03,280 --> 00:08:08,560
So that's kind of your large 
language model esque type of 

122
00:08:09,280 --> 00:08:11,360
ideas. 
You also then have as you kind 

123
00:08:11,360 --> 00:08:14,320
of mentioned there the this kind
of deep and that's kind of like 

124
00:08:14,320 --> 00:08:18,840
this deep learning. 
And what they try to do with 

125
00:08:18,840 --> 00:08:24,680
deep learning is that again it's
a subset of AI where they are 

126
00:08:24,680 --> 00:08:30,160
literally trying to kind of feed
and create this kind of neural 

127
00:08:30,160 --> 00:08:34,280
network that you know 
effectively learns by itself. 

128
00:08:35,120 --> 00:08:38,640
And so that's where chess 
element comes from. 

129
00:08:38,640 --> 00:08:41,280
So anything that kind of has 
deep blue, deep mind, anything 

130
00:08:41,280 --> 00:08:45,240
like that is kind of using this 
kind of deep learning kind of 

131
00:08:45,320 --> 00:08:50,160
process that uses artificial 
neural networks to try and get 

132
00:08:50,160 --> 00:08:54,720
it to comprehend and understand 
and attempt to reason. 

133
00:08:54,720 --> 00:08:58,880
But not always very well what 
they should and shouldn't be 

134
00:08:58,880 --> 00:09:01,280
doing at various points which 
wise that they can analyse all 

135
00:09:01,280 --> 00:09:05,640
that data and make those 
predictions because that's all 

136
00:09:05,640 --> 00:09:06,880
they are. 
And that's a kind of really 

137
00:09:06,880 --> 00:09:12,400
important thing to remember when
we're talking about AI, 

138
00:09:12,440 --> 00:09:19,360
particularly sort of generative 
AI, is that it's just a massive 

139
00:09:19,360 --> 00:09:22,640
prediction machine. 
And that has massive 

140
00:09:22,640 --> 00:09:27,640
ramifications for how you should
treat the kind of information 

141
00:09:27,640 --> 00:09:32,040
that it, you know, the outputs 
that it then provides you 

142
00:09:32,040 --> 00:09:36,760
because they're not necessarily 
searching the Internet for 

143
00:09:37,160 --> 00:09:40,760
accurate information. 
Everything that it creates is 

144
00:09:42,080 --> 00:09:46,200
new in the sense that it hasn't 
necessarily been written and 

145
00:09:46,200 --> 00:09:49,440
copyrighted previously. 
So. 

146
00:09:50,800 --> 00:09:53,840
But it can sound really 
convincing in the terms of what 

147
00:09:53,920 --> 00:09:56,960
you want it to give, which can 
make it seem like, Oh well, you 

148
00:09:56,960 --> 00:10:00,400
must have picked this up from 
the Internet or you know, I've 

149
00:10:00,400 --> 00:10:05,080
asked you for paragraph three, 
Chapter 4 and War and War and 

150
00:10:05,080 --> 00:10:07,160
Peace. 
And I want you to summarize 

151
00:10:07,160 --> 00:10:08,040
that. 
And then it kind of gives a 

152
00:10:08,040 --> 00:10:12,440
really what you would assume you
know, jurotish well, not 

153
00:10:12,440 --> 00:10:17,160
necessarily accurate plausible 
idea what that actually could 

154
00:10:17,160 --> 00:10:21,000
be, where it's an actual fact. 
You know, it sounds like it 

155
00:10:21,000 --> 00:10:24,720
could be summarizing it, but the
reality is, is that it's just 

156
00:10:25,640 --> 00:10:27,520
made something completely up. 
Yeah. 

157
00:10:27,520 --> 00:10:29,160
I mean, I think that's a really 
important distinction there. 

158
00:10:29,160 --> 00:10:33,160
Yeah, because I was almost 
conflating machine learning with

159
00:10:33,440 --> 00:10:36,400
artificial intelligence. 
That's not quite the case, no. 

160
00:10:36,800 --> 00:10:41,560
Machine learning is a a subset 
of AI. 

161
00:10:41,560 --> 00:10:43,520
It's a type of AI, but it's not 
the same thing. 

162
00:10:43,520 --> 00:10:48,640
And machine learning has been 
around for, yeah, ages on our 

163
00:10:48,640 --> 00:10:51,680
phone. 
You know, machine learning is 

164
00:10:51,680 --> 00:10:56,240
the kind of stuff that like, I 
don't know if you you're 

165
00:10:56,240 --> 00:10:59,240
probably, you know, far much, 
far, much more of a life than I 

166
00:10:59,240 --> 00:11:00,880
do. 
But I like making sure that if I

167
00:11:00,880 --> 00:11:04,440
take photos of people that my 
phone like, tells me who those 

168
00:11:04,440 --> 00:11:05,720
people are. 
So if I wanted to find all the 

169
00:11:05,720 --> 00:11:09,680
photos that I've had of you that
are in my phone, I could just 

170
00:11:09,680 --> 00:11:12,840
search for Kieran Mackel and it 
would kind of show me all the 

171
00:11:12,840 --> 00:11:15,440
photos that I that it thinks is 
you. 

172
00:11:15,440 --> 00:11:18,880
Now you kind of start training 
that model, then eventually it 

173
00:11:18,880 --> 00:11:20,680
kind of gets to learn who you 
are. 

174
00:11:21,480 --> 00:11:24,000
And so, you know, every now and 
then it just pops up saying, 

175
00:11:24,000 --> 00:11:27,360
hey, here's, you know, 5 new 
photos of of you and Kieran 

176
00:11:27,360 --> 00:11:29,040
Mackel. 
That's kind of like machine 

177
00:11:29,040 --> 00:11:31,840
learning. 
That's it kind of learning from 

178
00:11:32,160 --> 00:11:34,400
that data that you kind of feed 
it to kind of get to know you 

179
00:11:34,400 --> 00:11:39,880
and say the phone, iPhones, I'm 
sure Google are doing the same 

180
00:11:39,880 --> 00:11:41,240
as well. 
You know that kind of feature 

181
00:11:41,240 --> 00:11:44,880
has been there for for ages. 
So, so yeah, machine learning, 

182
00:11:44,880 --> 00:11:49,240
nothing kind of new, no, it's 
not something that's just brand 

183
00:11:49,240 --> 00:11:52,600
new to market. 
Such the the newish side of it 

184
00:11:52,600 --> 00:11:56,840
is this idea if it's been these 
generative AI tools. 

185
00:11:56,840 --> 00:12:00,680
This idea of you can prompt it 
with various things and then it 

186
00:12:00,680 --> 00:12:04,120
kind of gives you an output 
which could either be a text 

187
00:12:04,720 --> 00:12:08,200
output or an image output 
depending on which AI tool you 

188
00:12:08,200 --> 00:12:10,640
are choosing to use. 
My phone does that sometimes. 

189
00:12:10,640 --> 00:12:12,840
It does it for me, or it feels 
like it's done it for me. 

190
00:12:13,240 --> 00:12:16,680
I've actually been training mine
to identify me as someone else 

191
00:12:16,680 --> 00:12:20,240
in case I ever need to be 
extracted, and then they'll have

192
00:12:20,240 --> 00:12:22,120
my records. 
No Who's this guy? 

193
00:12:26,280 --> 00:12:30,280
Is artificial intelligence 
likely to change the way the 

194
00:12:30,280 --> 00:12:33,040
teachers work? 
I know this will be a surprising

195
00:12:33,040 --> 00:12:35,000
answer or not, but I actually 
don't think it's going to 

196
00:12:35,000 --> 00:12:37,960
necessarily fundamentally change
the way that we work. 

197
00:12:38,000 --> 00:12:41,360
I think it's just going to 
support teachers and how they do

198
00:12:41,360 --> 00:12:43,240
the work that they currently 
need to do. 

199
00:12:44,000 --> 00:12:51,600
I don't think we're a whole 
myriad of reasons that you know,

200
00:12:51,600 --> 00:12:54,360
we're going to automate a lot of
the things that we would like to

201
00:12:54,360 --> 00:13:00,280
automate from a data GDPR 
perspective, from a cost 

202
00:13:00,280 --> 00:13:04,040
perspective. 
I don't think it's kind of with 

203
00:13:04,240 --> 00:13:07,240
anywhere near that yet. 
But what I do think it will do 

204
00:13:07,880 --> 00:13:16,120
is kind of, you know, be the 
sort of digital teaching 

205
00:13:17,120 --> 00:13:20,160
assistant to the teacher, I 
can't say teaching assistant, 

206
00:13:20,160 --> 00:13:21,640
otherwise we might get confused 
with what we mean. 

207
00:13:21,640 --> 00:13:23,720
So it's definitely assistant to 
the teacher. 

208
00:13:25,280 --> 00:13:27,240
I kind of think the way that it 
kind of the best way that's 

209
00:13:27,240 --> 00:13:30,480
going to kind of help is if you 
think about the Pareto 

210
00:13:30,640 --> 00:13:35,080
principle, so you know 20% of 
the efforts deliver 80% of the 

211
00:13:35,080 --> 00:13:37,720
results. 
Like the hardest place to start 

212
00:13:37,720 --> 00:13:41,840
is obviously that first like 20%
of that, 20% like when you're 

213
00:13:41,840 --> 00:13:44,720
there with like a blank sheet of
paper because you have to do a 

214
00:13:44,800 --> 00:13:48,560
fill in a form or you have to 
sort of write a risk assessment 

215
00:13:48,560 --> 00:13:52,840
for a trip or you know, plan a 
lesson and you're just there 

216
00:13:52,840 --> 00:13:55,120
with that kind of blank sheet of
paper for ideas. 

217
00:13:55,520 --> 00:13:58,520
You know that first 20% of that 
20% is really hard. 

218
00:13:58,520 --> 00:14:02,520
And I kind of think that what 
generative AI can kind of really

219
00:14:02,520 --> 00:14:06,800
help us with is to kind of 
provide that launching point to 

220
00:14:06,800 --> 00:14:13,000
kind of help us with that first 
20% of the 20% that delivers the

221
00:14:13,000 --> 00:14:16,720
80% of the output, if that makes
sense, which is the hardest 

222
00:14:16,720 --> 00:14:18,720
part. 
But just by giving us that kind 

223
00:14:18,720 --> 00:14:22,200
of launchpad, get these ideas 
flowing, I think it's going to 

224
00:14:22,200 --> 00:14:24,680
be really, you know, quite 
useful in that regard to 

225
00:14:24,680 --> 00:14:27,840
teachers because they say it 
kind of gives them that that 

226
00:14:27,840 --> 00:14:32,120
first could be that first step 
or, you know, just, you know, a 

227
00:14:32,120 --> 00:14:35,960
couple of ideas that could, you 
know, be what they need to then,

228
00:14:35,960 --> 00:14:39,520
you know, get into a state of 
flow so that they can kind of 

229
00:14:40,120 --> 00:14:43,160
finish off whatever task and 
activity that they need to do. 

230
00:14:43,720 --> 00:14:48,600
So yeah, I don't think it's 
going to massively change what 

231
00:14:48,600 --> 00:14:52,920
teachers do, but I think it will
change sort of how teachers are 

232
00:14:52,920 --> 00:15:01,400
going to do it, particularly I 
think with things like and again

233
00:15:01,400 --> 00:15:07,840
depends on budgets and data 
protection and things like that.

234
00:15:07,840 --> 00:15:12,760
But Google services they're 
starting to integrate Bard which

235
00:15:12,760 --> 00:15:15,160
is Google's large language 
model. 

236
00:15:16,960 --> 00:15:24,520
Microsoft it has Co pilots which
is their large language model 

237
00:15:25,800 --> 00:15:29,640
and all these and this kind of 
assistant we want to a better 

238
00:15:29,640 --> 00:15:31,760
word can talk to your office 
apps. 

239
00:15:31,760 --> 00:15:37,640
So Co pilot can talk to outlook,
word, PowerPoint, Onenotes 

240
00:15:38,080 --> 00:15:41,720
whatever it may be. 
But the same for the Google 

241
00:15:41,720 --> 00:15:45,040
equivalents. 
And So what what would really do

242
00:15:45,040 --> 00:15:52,160
is kind of really support 
teachers bringing all the 

243
00:15:52,160 --> 00:15:56,400
information that currently sits 
separately within those 

244
00:15:56,400 --> 00:15:59,320
individual applications and 
bring them together and being 

245
00:15:59,320 --> 00:16:02,720
able to kind of compile that 
information together and you 

246
00:16:02,720 --> 00:16:04,120
know, produce some sort of 
output. 

247
00:16:04,120 --> 00:16:11,640
So you might have for example 
teacher planning a let's stick 

248
00:16:11,640 --> 00:16:18,300
with planning a school trip. 
So she might want to ask Co 

249
00:16:18,300 --> 00:16:23,320
pilot, I want you to bring up or
kind of show me everything that 

250
00:16:23,320 --> 00:16:28,960
I've done so far to support the 
school trip to the Tower of 

251
00:16:28,960 --> 00:16:31,120
London. 
And so the e-mail that she sent 

252
00:16:31,120 --> 00:16:35,000
to the to SLT asking for the 
trip would come up 

253
00:16:35,000 --> 00:16:39,640
automatically. 
The, the Word document that she 

254
00:16:40,240 --> 00:16:43,640
put together for the letter 
would come up the PowerPoint 

255
00:16:43,640 --> 00:16:47,200
presentation that she's done for
the parents to explain the 

256
00:16:47,200 --> 00:16:50,400
purpose of the trip and you know
how they can support and all 

257
00:16:50,400 --> 00:16:52,880
that stuff will come up. 
And obviously then you can kind 

258
00:16:52,880 --> 00:16:57,000
of, you know, combine all of 
that data that's in those 

259
00:16:57,000 --> 00:17:00,880
particular documents, you know 
to potentially then create 

260
00:17:00,880 --> 00:17:02,520
something, you know quite 
powerful. 

261
00:17:02,520 --> 00:17:07,520
And so I do feel that side of 
it, that sort of digital admin 

262
00:17:07,560 --> 00:17:12,520
assistant, you know, that could 
actually probably, you know, 

263
00:17:13,079 --> 00:17:16,200
again not change the way 
teachers work, but change how 

264
00:17:16,200 --> 00:17:18,640
teachers need to do the work 
that they need to do. 

265
00:17:20,560 --> 00:17:23,359
However, as I mentioned 
previously, there's the whole 

266
00:17:23,359 --> 00:17:25,880
thing about, well, how much this
does is this going to cost 

267
00:17:25,880 --> 00:17:28,480
schools extra. 
And obviously as we're talking 

268
00:17:28,480 --> 00:17:32,360
about, you know, young children 
here, you know, you need to give

269
00:17:32,520 --> 00:17:38,440
Co pilots access to effectively 
your Microsoft account. 

270
00:17:39,440 --> 00:17:42,640
And so if you are a complete 
Microsoft school, that is all 

271
00:17:42,640 --> 00:17:48,120
the data that you may hold on 
children and whatever else 

272
00:17:48,840 --> 00:17:50,720
within those different 
applications. 

273
00:17:51,560 --> 00:17:57,000
Those then need to go off to a 
Microsoft server or you know Co 

274
00:17:57,000 --> 00:17:58,840
pilot to do the things that it 
needs to do. 

275
00:18:00,360 --> 00:18:02,400
And obviously you know there are
massive implications and if 

276
00:18:02,400 --> 00:18:07,120
anything you know does go awry. 
But that is definitely I think 

277
00:18:07,120 --> 00:18:11,080
one way that I think will save 
teachers time from that kind of 

278
00:18:11,080 --> 00:18:16,120
admin side of it and that kind 
of aggregation of all this data 

279
00:18:16,120 --> 00:18:19,240
that teachers accumulate or 
maybe it's just me and I just 

280
00:18:19,240 --> 00:18:21,440
accumulate lots of emails about 
different things. 

281
00:18:21,440 --> 00:18:25,160
And you know my filing system of
organising things has never been

282
00:18:25,800 --> 00:18:29,960
as I would like it to be. 
I'm I'm no Adam Smith with his 

283
00:18:29,960 --> 00:18:34,320
notion and things like that. 
But you know, I think it could 

284
00:18:34,320 --> 00:18:37,160
help teachers with that regard 
and I think that's not the 

285
00:18:37,160 --> 00:18:38,400
reason teachers come into 
school. 

286
00:18:38,400 --> 00:18:42,200
So anything that helps them to 
that side of it, which means 

287
00:18:42,200 --> 00:18:43,800
that they can focus more on 
their teaching. 

288
00:18:44,960 --> 00:18:46,640
I don't want to go as far as 
saying that there's probably 

289
00:18:46,640 --> 00:18:49,960
like a moral reason there that 
schools should be looking into 

290
00:18:49,960 --> 00:18:53,800
this because by going down that 
Ave. they're opening up more 

291
00:18:53,800 --> 00:19:01,040
kind of ethical issues. 
But I definitely think exploring

292
00:19:01,040 --> 00:19:03,840
such avenues, particularly in 
this current climate, is 

293
00:19:03,840 --> 00:19:07,480
something that, you know, 
schools, large trusts, the DfE, 

294
00:19:07,480 --> 00:19:10,520
should, should be considering. 
I think it sounds really 

295
00:19:10,520 --> 00:19:12,520
interesting. 
I would love that as a teacher 

296
00:19:12,960 --> 00:19:16,600
to have something that would 
help me structure my personal 

297
00:19:16,600 --> 00:19:20,560
organization. 
A lot of the a lot of the early 

298
00:19:20,560 --> 00:19:27,560
conversations about artificial 
intelligence seems to allude to 

299
00:19:27,560 --> 00:19:32,520
the outsourcing of thinking. 
And obviously, the main theme of

300
00:19:32,520 --> 00:19:35,120
the podcast is that thinking is 
the most important thing 

301
00:19:35,200 --> 00:19:38,800
teachers can do. 
Are you saying that there's no 

302
00:19:38,800 --> 00:19:43,560
way that artificial intelligence
will ever take the thinking art 

303
00:19:43,560 --> 00:19:47,120
of teaching? 
You know and and you know, Hey, 

304
00:19:47,560 --> 00:19:54,000
Bard, I want a Lesson plan on, I
don't know, rock and Romans. 

305
00:19:54,680 --> 00:19:55,880
Is that that day ever going to 
come? 

306
00:19:56,080 --> 00:19:59,000
Yeah, that day will come. 
But I don't necessarily classify

307
00:19:59,000 --> 00:20:03,080
that as the AI doing the the 
thinking for you. 

308
00:20:04,600 --> 00:20:09,520
As I say, you know, I would be 
amazed if anyone takes that, you

309
00:20:09,520 --> 00:20:15,040
know, puts that into a 
generative AI tool and, you 

310
00:20:15,200 --> 00:20:17,520
know, doesn't then do anything 
with it. 

311
00:20:18,960 --> 00:20:21,480
As I say, you know, it starts 
off the thinking process. 

312
00:20:21,480 --> 00:20:23,920
It gives you some ideas to to 
think about but you know 

313
00:20:24,040 --> 00:20:30,760
fundamentally you know you then 
have to I think with that 

314
00:20:30,760 --> 00:20:34,120
information again the 1st port 
of port of call and it kind of 

315
00:20:34,120 --> 00:20:37,440
comes back to something we'll 
discuss later on is you know you

316
00:20:37,440 --> 00:20:39,800
have to check for the accuracy 
if you're using a large language

317
00:20:39,800 --> 00:20:42,760
model to do this. 
You know you have to check for 

318
00:20:42,760 --> 00:20:43,920
the accuracy. 
So if you. 

319
00:20:44,360 --> 00:20:52,000
So my preferred way of doing it 
would be that you have, you 

320
00:20:52,040 --> 00:20:56,800
know, quite a tight curriculum 
in the terms that you're quite 

321
00:20:56,800 --> 00:20:58,720
explicit as to what want 
students to know. 

322
00:20:58,720 --> 00:21:04,800
So let's say keeping on the 
theme of rocking Romans. 

323
00:21:04,800 --> 00:21:08,440
So you know, Caesar tries to 
invade in 55, he tried again in 

324
00:21:08,440 --> 00:21:13,680
54 and he fails because I think 
the weather was particularly 

325
00:21:13,680 --> 00:21:17,640
bad, something like that in one 
of the invasions. 

326
00:21:17,640 --> 00:21:22,120
And the the Celtic tribes were 
able to kind of have a bit of a 

327
00:21:22,120 --> 00:21:24,720
ceasefire between themselves to 
come together to defend the 

328
00:21:24,720 --> 00:21:26,280
Romans. 
Let's say that's what you wanted

329
00:21:26,280 --> 00:21:30,640
out of that first lesson. 
If you told the AI that you 

330
00:21:30,640 --> 00:21:31,960
know, these are the parameters 
of my lesson. 

331
00:21:31,960 --> 00:21:32,920
This is what I want kids to 
know. 

332
00:21:32,920 --> 00:21:36,240
I want them to know that, you 
know, Julius Caesar invaded 55, 

333
00:21:36,240 --> 00:21:42,000
then 54 and then it failed. 
One of the reasons why the 54 

334
00:21:42,200 --> 00:21:45,840
invasion failed or two reasons 
why the 54 invasion failed was 

335
00:21:45,840 --> 00:21:49,640
because the weather was against 
him. 

336
00:21:49,640 --> 00:21:51,880
So he had to invade later than 
he originally planned. 

337
00:21:52,000 --> 00:21:56,320
And the Celtic tribes of Britain
at the time, you know, came 

338
00:21:56,320 --> 00:22:00,280
together to, you know, stand 
against him, make me a Lesson 

339
00:22:00,280 --> 00:22:03,680
plan from that is going to be 
far more beneficial, far more 

340
00:22:03,680 --> 00:22:09,280
effective, far more accurate 
because you, the teacher are 

341
00:22:09,280 --> 00:22:13,600
kind of, you know, narrowing 
those parameters for which this 

342
00:22:13,600 --> 00:22:17,360
predictive model is going to be 
working from. 

343
00:22:18,760 --> 00:22:21,720
So the output that it gives you 
is going to be far more accurate

344
00:22:21,720 --> 00:22:26,640
to what you need versus 
something like hi ChatGPT. 

345
00:22:26,640 --> 00:22:29,480
I'm teaching a lesson on the 
Romans. 

346
00:22:30,080 --> 00:22:33,920
It's for year three. 
You know, hit the button, go, 

347
00:22:34,480 --> 00:22:37,440
because you know that's just 
going to deliver all sorts of 

348
00:22:38,320 --> 00:22:41,200
weird and wonderful things with 
potential inaccuracies. 

349
00:22:41,200 --> 00:22:48,000
So the more thinking that you do
actually before you ask what you

350
00:22:48,000 --> 00:22:53,040
want, the more useful the that 
output is going to be in sort of

351
00:22:53,040 --> 00:22:55,680
a big. 
And that's kind of partly the 

352
00:22:55,680 --> 00:23:01,040
reason why I've gone down this 
kind of like this prompt Ave. 

353
00:23:01,680 --> 00:23:05,720
because I kind of really want to
make people aware that this AI 

354
00:23:07,080 --> 00:23:10,920
tools and technology is can be 
really effective. 

355
00:23:11,480 --> 00:23:15,080
But it really is sort of that 
intersection between teacher 

356
00:23:15,080 --> 00:23:19,400
expertise and your knowledge of 
how to provide a really useful 

357
00:23:19,400 --> 00:23:23,160
prompt. 
It's not an outsourcing of 

358
00:23:23,160 --> 00:23:24,880
teacher expertise. 
Far from it. 

359
00:23:25,280 --> 00:23:26,640
Now will we ever get there 
though? 

360
00:23:27,400 --> 00:23:32,200
I mean, you know, who knows? 
But right now we can't. 

361
00:23:32,200 --> 00:23:35,440
We shouldn't. 
I I was listening to a podcast 

362
00:23:35,440 --> 00:23:37,480
and it was from a totally 
different field with the Talmud 

363
00:23:37,480 --> 00:23:45,800
AI and they were saying that 
between now and April next year 

364
00:23:46,400 --> 00:23:48,200
there's going to be a massive 
explosion. 

365
00:23:48,240 --> 00:23:49,720
You know, if we think about 
what? 

366
00:23:51,280 --> 00:23:54,200
The explosion that happened last
October when the world became 

367
00:23:54,200 --> 00:23:59,320
aware of this large language and
model, and then the 12 things 

368
00:23:59,320 --> 00:24:01,320
that have happened in 12 months 
since then, But I think this 

369
00:24:01,320 --> 00:24:04,360
next, what, six months will be 
massive. 

370
00:24:04,360 --> 00:24:07,120
So who knows what will be 
possible. 

371
00:24:07,120 --> 00:24:10,200
But I really like the idea that 
it's going to take a lot of 

372
00:24:10,200 --> 00:24:14,120
expertise in two domains now to 
actually get the most from it. 

373
00:24:14,120 --> 00:24:17,040
And you can't just type in. 
I'm teaching fractions. 

374
00:24:17,360 --> 00:24:19,000
Off you go. 
You know, there's my fractions 

375
00:24:19,000 --> 00:24:22,600
plan, you know that. 
That gives me some reassurance. 

376
00:24:22,800 --> 00:24:26,520
We're recording this on Monday 
the 30th of October 2023. 

377
00:24:26,960 --> 00:24:30,440
You know, this is probably going
to be out of date before even. 

378
00:24:30,440 --> 00:24:32,000
I don't know when you're 
planning on putting this one out

379
00:24:32,000 --> 00:24:34,760
there, but I would go likely 
before where this gets put out 

380
00:24:34,760 --> 00:24:39,520
to to listeners and elements of 
what I've said is probably going

381
00:24:39,520 --> 00:24:42,120
to be out of date already, which
is just goes to show how fast 

382
00:24:42,120 --> 00:24:45,480
moving you know this field 
really is. 

383
00:24:46,840 --> 00:24:50,240
Yeah, absolutely. 
You mentioned prompts, so you're

384
00:24:50,240 --> 00:24:57,520
alluding to the newsletter that 
you have started to release. 

385
00:24:57,880 --> 00:25:04,680
So myself, past participant on 
the show, Pats McRae, who I'm 

386
00:25:04,680 --> 00:25:12,640
sure people will know from all 
his concise books, which made me

387
00:25:12,640 --> 00:25:14,440
think that perhaps they were 
mostly written by AI. 

388
00:25:14,520 --> 00:25:17,600
I'm not sure yet. 
I haven't had the guts to to ask

389
00:25:17,600 --> 00:25:20,320
him yet whether he was actually 
just using large language models

390
00:25:20,320 --> 00:25:25,000
to write most of those, but I'm 
sure I'll ask him at some point.

391
00:25:25,280 --> 00:25:29,160
We kind of had this idea of 
putting this out there to help 

392
00:25:29,160 --> 00:25:36,360
teachers navigate the the murky 
waters of these generative AI 

393
00:25:38,360 --> 00:25:41,320
tools that are out there. 
Obviously we alluded to earlier 

394
00:25:41,320 --> 00:25:43,920
the kind of the kind of Twitter 
bots that are appearing here 

395
00:25:43,920 --> 00:25:47,280
there and everywhere and other 
kind of various other you know 

396
00:25:47,520 --> 00:25:52,800
AI tools that are marketed 
towards people using and seeing 

397
00:25:52,800 --> 00:26:00,520
if we can provide a different 
Ave. of support which I say it's

398
00:26:00,520 --> 00:26:04,880
predominantly through this idea 
of of prompting which is where 

399
00:26:04,960 --> 00:26:11,760
you most of these generative AI 
tools like ChatGPT and Dolly, 

400
00:26:11,760 --> 00:26:15,360
Mid Journey, whatever they may 
be predominantly require on the 

401
00:26:15,360 --> 00:26:18,200
text inputs. 
And that's called your prompt. 

402
00:26:18,320 --> 00:26:25,000
And effectively the this entire 
sort of field has arisen 

403
00:26:25,000 --> 00:26:30,800
recently called prompt 
engineering which is effectively

404
00:26:30,800 --> 00:26:34,480
I'm really starting to not like 
the term, but I'll use it 

405
00:26:34,480 --> 00:26:35,800
because I can't think of another
one. 

406
00:26:36,120 --> 00:26:40,840
But it's effectively the science
of how do we create effective 

407
00:26:40,840 --> 00:26:44,480
prompts to make sure that the 
output that we then get from 

408
00:26:44,480 --> 00:26:50,840
this generative AI model is what
we envision that prompt to be. 

409
00:26:51,440 --> 00:26:56,280
So as I said the these are 
effectively prediction machines.

410
00:26:56,280 --> 00:26:59,000
So if we are really vague, you 
know, you say our one Lesson 

411
00:26:59,000 --> 00:27:05,280
plan on the Romans, you might 
well get a Lesson plan back that

412
00:27:05,440 --> 00:27:10,040
includes things like this is how
you should support different 

413
00:27:10,040 --> 00:27:14,240
learning styles, for example, 
which evidently, you know, we 

414
00:27:14,240 --> 00:27:18,160
know is not a good use of our 
time, but clearly within the the

415
00:27:18,160 --> 00:27:21,520
trained data sets it has 
information about learning 

416
00:27:21,520 --> 00:27:23,120
styles. 
And so because we were quite 

417
00:27:23,120 --> 00:27:28,160
vague with our prompt, we, you 
know, it gave us back something.

418
00:27:28,160 --> 00:27:32,520
So the more we can kind of find 
the underlying structures and 

419
00:27:32,520 --> 00:27:34,640
play around with those 
underlying structures of what 

420
00:27:34,640 --> 00:27:39,680
effective prompting is, the 
higher quality, the output 

421
00:27:41,440 --> 00:27:43,960
effectively and hopefully the 
more useful than it then is for 

422
00:27:43,960 --> 00:27:48,840
teachers to to use in upcoming 
lessons or admin for example. 

423
00:27:49,720 --> 00:27:53,280
And so yes, one of the way that 
we started this process is 

424
00:27:53,280 --> 00:27:56,760
through our our weekly 
newsletter, Teacher Prompts, 

425
00:27:56,760 --> 00:28:01,600
which is there to kind of 
upskill teachers not just in 

426
00:28:01,600 --> 00:28:06,240
sort of prompts, although it we 
try to make sure that it does do

427
00:28:06,240 --> 00:28:09,960
that. 
It's kind of there just to give 

428
00:28:11,080 --> 00:28:16,120
dominantly, you know, educators,
you know, up to date news in the

429
00:28:17,720 --> 00:28:21,240
AI world that sort of where 
possible sort of directly 

430
00:28:21,240 --> 00:28:27,360
influences education. 
So for example, at 2 bits of 

431
00:28:27,360 --> 00:28:32,480
news that will probably make it 
to tomorrow's newsletter is that

432
00:28:33,480 --> 00:28:36,080
you know the government's just 
recently handed over Oak 2 

433
00:28:36,080 --> 00:28:41,560
million to explore AI use there.
I've had a bit of a play with 

434
00:28:41,560 --> 00:28:44,280
some of these tools already and 
you know they need work. 

435
00:28:44,280 --> 00:28:49,160
But I'm sure you know the more 
this is in front of teachers, 

436
00:28:49,160 --> 00:28:50,800
the more useful it's going to 
be. 

437
00:28:51,520 --> 00:28:59,720
And you can now upload if you 
purchase ChatGPT Plus to access 

438
00:28:59,720 --> 00:29:05,880
the the GPT 4 model, which is 
you know, has more parameters, 

439
00:29:05,880 --> 00:29:12,560
is trained on more data, more 
accurate, you can sort of now 

440
00:29:12,560 --> 00:29:18,080
drop PDFs and other kind of 
files into it and it can then 

441
00:29:18,120 --> 00:29:21,760
you know, summarize and you can 
ask questions and interrogate it

442
00:29:21,760 --> 00:29:24,160
about that particular file that 
you've dropped in, et cetera, et

443
00:29:24,160 --> 00:29:26,760
cetera. 
So and then we go in about you 

444
00:29:26,760 --> 00:29:30,160
know what effective, you know, 
some effective prompts that will

445
00:29:30,160 --> 00:29:32,840
be useful for for teachers. 
Yeah. 

446
00:29:32,840 --> 00:29:36,400
So every week and I like that 
you always have you know in this

447
00:29:36,400 --> 00:29:42,240
in this I don't know edition in 
this this one episode you you 

448
00:29:42,240 --> 00:29:45,640
will find and then you get like 
a couple of links and you've got

449
00:29:45,720 --> 00:29:50,680
and this image yeah always try 
throwing a generative AI image 

450
00:29:50,680 --> 00:29:53,160
there just to they're quite 
good. 

451
00:29:53,160 --> 00:29:55,600
I mean, you know, we had some 
good fun making some of the 

452
00:29:55,600 --> 00:30:00,000
other day, didn't we, in our 
little little group chat of the 

453
00:30:00,000 --> 00:30:03,680
weird and wonderful things that 
you could you could do with 

454
00:30:03,680 --> 00:30:06,840
that. 
I really enjoyed the the UP 

455
00:30:06,840 --> 00:30:12,240
sequel down. 
But it was funny because you 

456
00:30:12,240 --> 00:30:14,480
immediately, you could 
immediately get that sense that 

457
00:30:14,480 --> 00:30:18,360
this was UP If this was, you 
know, it was all themed on the 

458
00:30:18,360 --> 00:30:22,720
idea of UP. 
And, you know, I'm not going to 

459
00:30:22,720 --> 00:30:25,800
say what's, you know, holding me
back from creating high quality 

460
00:30:25,800 --> 00:30:31,760
educational resources has been 
my poor graphic design ability. 

461
00:30:31,760 --> 00:30:36,840
But I'm far more inclined to 
give it a go now that I know I 

462
00:30:36,840 --> 00:30:44,640
can just prompt Dolly to give me
a few, you know, useful images 

463
00:30:44,640 --> 00:30:46,560
to use. 
So yeah, it's going to be 

464
00:30:46,560 --> 00:30:48,840
useful. 
Yeah, I I mean, it does take 

465
00:30:48,840 --> 00:30:51,120
some prompting because I gave it
some lazy prompts and got some 

466
00:30:51,120 --> 00:30:53,520
nonsense back. 
My best one were The The 

467
00:30:53,520 --> 00:30:55,680
Beatles, but they're a band 
called The Beatles, but they're 

468
00:30:55,680 --> 00:30:58,840
actually Beatles. 
And then dishwasher has fallen. 

469
00:30:59,720 --> 00:31:01,480
Dishwasher has fallen. 
It was a great one. 

470
00:31:01,760 --> 00:31:04,760
They will put these when the 
episode goes out. 

471
00:31:04,760 --> 00:31:06,640
We'll find them and people. 
Can have link? 

472
00:31:06,840 --> 00:31:08,480
Yeah, good guy. 
Yeah. 

473
00:31:08,480 --> 00:31:10,320
I think it's season what we're 
using. 

474
00:31:10,320 --> 00:31:12,760
We're using Bards image 
generator, so it's called. 

475
00:31:13,120 --> 00:31:18,160
We were using Bing. 
Bing, not Bard. 

476
00:31:18,160 --> 00:31:20,600
Yeah, Bard is Google Bing, Bing 
is. 

477
00:31:20,600 --> 00:31:24,520
Microsoft Yeah, that's the 
search engine that one uses, 

478
00:31:24,920 --> 00:31:32,160
exactly. 
So technology and education have

479
00:31:32,160 --> 00:31:36,680
a very interesting relationship.
Sometimes millions of pounds can

480
00:31:36,680 --> 00:31:40,240
be spent in ways that were 
perhaps ill advised. 

481
00:31:41,120 --> 00:31:45,240
What are the chances that this 
AI revolution is smoke and 

482
00:31:45,240 --> 00:31:49,320
mirrors like those other aspects
of technology that have proven 

483
00:31:49,760 --> 00:31:51,800
not to be as effective in AI? 
Don't know. 

484
00:31:52,720 --> 00:31:55,840
Changing our practice or our 
ways of working, I mean there's 

485
00:31:55,840 --> 00:31:57,920
always the chances. 
Now you know we've seen it with,

486
00:31:58,000 --> 00:32:00,160
you know, interactive 
whiteboards were not the 

487
00:32:00,560 --> 00:32:04,560
solution that we were sold. 
They would be 1 to one devices 

488
00:32:04,560 --> 00:32:08,760
and iPads certainly haven't been
the thing that they've done. 

489
00:32:09,000 --> 00:32:14,200
The reason I feel like this is 
different one because you don't 

490
00:32:14,200 --> 00:32:20,360
necessarily need almost there's 
a continuum here. 

491
00:32:20,600 --> 00:32:24,400
You know if you were to go as I 
mentioned you know the whole 

492
00:32:25,000 --> 00:32:29,400
virtual admin assistant via Bard
or Co pilot and that's something

493
00:32:29,400 --> 00:32:32,360
that your school purchased as 
part of you know, the Microsoft 

494
00:32:32,360 --> 00:32:34,880
license, then yes, obviously 
there is a likely to be an 

495
00:32:34,880 --> 00:32:39,160
additional cost to that. 
But you can get in there, you 

496
00:32:39,160 --> 00:32:42,920
know, without having to without 
price being a barrier. 

497
00:32:42,920 --> 00:32:46,920
You know ChatGPT 3.5 is free to 
use. 

498
00:32:48,360 --> 00:32:51,880
Bing is free to use as well. 
All you have to do is just make 

499
00:32:51,880 --> 00:32:55,280
accounts for these things and I 
feel as well you know the one to

500
00:32:55,280 --> 00:32:59,960
one iPads or you know one to one
devices, the interactive 

501
00:32:59,960 --> 00:33:06,200
whiteboards, all of that stuff 
that was very much student 

502
00:33:06,360 --> 00:33:11,400
facing uses of technology, this 
sort of generative AI stuff. 

503
00:33:13,360 --> 00:33:18,240
Yes you might put resources that
you've created through this in 

504
00:33:18,240 --> 00:33:21,720
front of the pupils but it is 
very much all happening sort of 

505
00:33:21,720 --> 00:33:24,760
you know in the background 
backstage as you were. 

506
00:33:25,160 --> 00:33:26,840
You know it's the planning 
processes. 

507
00:33:26,840 --> 00:33:30,120
It's the that you know that deep
thinking that we're doing before

508
00:33:30,120 --> 00:33:31,760
we get in front of the 
classroom. 

509
00:33:32,760 --> 00:33:37,960
And because of that element I 
think it will be different I 

510
00:33:37,960 --> 00:33:42,040
think because I mean the the 
irony is you know you know I'm 

511
00:33:43,640 --> 00:33:48,640
I'm I'm no fan of you know 
Soonak as you know. 

512
00:33:48,640 --> 00:33:51,240
But I think we all laughed a 
couple of years ago when he said

513
00:33:51,240 --> 00:33:55,560
he was going to plough a bit of 
money into AI and we all sort of

514
00:33:56,400 --> 00:34:01,120
can you know laughed at him. 
But you know I think you know 

515
00:34:02,480 --> 00:34:06,040
credit where credit is due. 
You know, he was potentially 

516
00:34:06,040 --> 00:34:09,280
right in that, you know, this is
something that is going to 

517
00:34:09,280 --> 00:34:13,320
really potentially, you know, 
support teachers. 

518
00:34:13,320 --> 00:34:17,040
And I think, you know, with 
their putting their money where 

519
00:34:17,040 --> 00:34:22,320
their mouth is to a degree by 
giving you know, Oak Nationals 

520
00:34:22,320 --> 00:34:24,760
some funding to kind of explore 
this. 

521
00:34:24,760 --> 00:34:28,800
And Oak National are putting all
their tools out there, which is 

522
00:34:29,639 --> 00:34:32,080
good and interesting. 
So, you know, teachers are going

523
00:34:32,080 --> 00:34:36,800
to, you know, see this. 
I'm sure there are teachers who 

524
00:34:36,800 --> 00:34:40,920
still don't know how to access 
these things and that these 

525
00:34:40,920 --> 00:34:42,239
things exist. 
Absolutely. 

526
00:34:42,239 --> 00:34:45,440
But those teachers that know 
about interactive whiteboards, 

527
00:34:45,440 --> 00:34:50,280
those teachers that know about 
large language models and 

528
00:34:50,280 --> 00:34:55,600
generative AII, bet they've 
played more with generative AI 

529
00:34:55,600 --> 00:35:00,080
and what that can do for them 
than they ever did about, you 

530
00:35:00,080 --> 00:35:02,920
know, let's have a look and see 
what what smart can do. 

531
00:35:03,480 --> 00:35:05,560
You know, they found infinite 
Clover and thought great, this 

532
00:35:05,560 --> 00:35:09,080
is my job done and I don't need,
don't need anything else. 

533
00:35:09,920 --> 00:35:12,160
So yeah, this feels kind of 
different because price isn't 

534
00:35:12,160 --> 00:35:16,360
necessarily a barrier to access.
There are three versions that 

535
00:35:16,360 --> 00:35:19,480
you can use. 
And because it's sort of, it's 

536
00:35:19,480 --> 00:35:22,920
not happening in the classroom, 
it's happening in that kind of 

537
00:35:22,920 --> 00:35:26,000
planning stage. 
You know, I think that there is 

538
00:35:26,680 --> 00:35:30,960
more scope for this to be 
successful than previous 

539
00:35:31,320 --> 00:35:33,800
elements of technology. 
But again, all it would take is 

540
00:35:33,800 --> 00:35:38,240
1 massive data breach, then the 
whole, you know, House of Cards 

541
00:35:38,240 --> 00:35:42,200
to come crumbling down somehow. 
So yeah, got to be careful with 

542
00:35:42,200 --> 00:35:44,400
it. 
I think it has more chance than 

543
00:35:44,840 --> 00:35:48,560
previous tech endeavours, but as
I say, it's certainly not 

544
00:35:48,560 --> 00:35:51,800
infallible by any means. 
If it's used in the way that you

545
00:35:51,800 --> 00:35:54,880
describe, it could be used. 
I think there's a fairly good 

546
00:35:54,880 --> 00:35:56,880
chance that it yeah, it could 
be. 

547
00:35:57,720 --> 00:35:59,680
There is I mean I think I 
mentioned this, I can't remember

548
00:35:59,680 --> 00:36:03,520
which episode it was. 
There is a quite a large. 

549
00:36:04,400 --> 00:36:06,680
I think it's a charity 
organization. 

550
00:36:07,280 --> 00:36:11,920
That's involved in the education
of other people and their 

551
00:36:11,920 --> 00:36:20,320
investment in AI has been trying
to get the the computer or the 

552
00:36:20,320 --> 00:36:26,640
program to give feedback on 
misconceptions and things like 

553
00:36:26,640 --> 00:36:31,760
that there. 
It feels like that's the kind of

554
00:36:31,760 --> 00:36:34,080
thing. 
I mean, you would need a whole 

555
00:36:34,080 --> 00:36:38,760
lot of information. 
To make an accurate prediction 

556
00:36:38,920 --> 00:36:44,600
about what the misconception was
or why the error had formed, Do 

557
00:36:44,600 --> 00:36:47,800
you reckon there's, you know, is
that indicative of how people 

558
00:36:47,800 --> 00:36:51,920
in, you know Silicon Valley at 
the minute maybe a little bit 

559
00:36:51,920 --> 00:36:55,920
further beyond, are are 
funneling their investment in 

560
00:36:55,920 --> 00:36:59,480
education and AII or do you 
think it is as as a sort of an 

561
00:36:59,480 --> 00:37:02,520
assistant to the regional 
manager? 

562
00:37:04,320 --> 00:37:07,400
That's a good question and what 
I haven't really thought about. 

563
00:37:07,640 --> 00:37:15,400
I would say the former purely on
no data or any you know, smart 

564
00:37:15,400 --> 00:37:18,840
hypothesis behind it other than 
you know in Silicon Valley, you 

565
00:37:18,840 --> 00:37:21,000
know data. 
Data is king. 

566
00:37:21,320 --> 00:37:25,360
There's that running joke where 
if you don't pay for a service, 

567
00:37:27,480 --> 00:37:30,800
you know you are you are the you
are the, you are the product, 

568
00:37:32,320 --> 00:37:36,600
Which is why I always pay for, 
you know, pay for Apple and try 

569
00:37:36,600 --> 00:37:38,160
to avoid Google wherever 
possible. 

570
00:37:38,600 --> 00:37:40,160
That's my biases coming in 
there. 

571
00:37:40,360 --> 00:37:43,080
And so the more data you have, 
the more valuable data is, the 

572
00:37:43,080 --> 00:37:45,760
more it can be used to target at
specific things. 

573
00:37:45,760 --> 00:37:50,440
And you know, we've got no idea 
how advertisers are somehow 

574
00:37:50,440 --> 00:37:55,360
going to find their mitts into 
all of this data and do stuff 

575
00:37:55,360 --> 00:38:00,680
with it to try to target more 
ads at you in a more kind of 

576
00:38:01,080 --> 00:38:05,400
specific fashion. 
But it's evidently, you know, in

577
00:38:05,400 --> 00:38:09,360
the past, it's paid off having 
as much data about as many 

578
00:38:09,360 --> 00:38:14,280
people as you can. 
So you can sort of generate, you

579
00:38:14,280 --> 00:38:19,280
know, rough profiles on who 
these people are, what they're 

580
00:38:19,280 --> 00:38:22,600
likely to be into, who they're 
likely to vote for and you know 

581
00:38:22,600 --> 00:38:27,440
all that sort of real semi 
conspiracy Cambridge Analytica 

582
00:38:27,440 --> 00:38:29,360
sort of stuff that you can kind 
of get into I. 

583
00:38:29,560 --> 00:38:33,040
Mean we don't have to legal back
in to navigate this storm that 

584
00:38:33,240 --> 00:38:37,240
is potentially brewing here. 
Perhaps we'll say yes, I I see 

585
00:38:37,240 --> 00:38:43,840
what you mean now because yeah. 
And lots of tech geniuses try to

586
00:38:43,840 --> 00:38:46,640
get involved in education in 
different ways to varying 

587
00:38:46,640 --> 00:38:49,960
degrees of of success. 
So it'll be interesting to see 

588
00:38:49,960 --> 00:38:52,280
how they go. 
I mean, what's really clear? 

589
00:38:52,800 --> 00:38:55,520
Is that over the last 12 months,
you've spent a lot of time 

590
00:38:56,400 --> 00:38:59,400
exploring this field and getting
to know this field because the, 

591
00:38:59,480 --> 00:39:03,960
you know, the, the quality of 
your answers today absolutely 

592
00:39:03,960 --> 00:39:05,960
blown my mind. 
Well, maybe maybe it speaks to 

593
00:39:05,960 --> 00:39:09,880
how you little I've properly 
engaged with this, but you know,

594
00:39:09,880 --> 00:39:12,520
I've learned a lot about how 
this actually might be. 

595
00:39:12,520 --> 00:39:17,520
You know, my skepticism is being
eroded gradually and with every 

596
00:39:17,520 --> 00:39:19,840
response. 
And so it's really clear you 

597
00:39:19,840 --> 00:39:21,480
spend a lot of time thinking 
about this. 

598
00:39:22,240 --> 00:39:23,280
But how we can get the most of 
it? 

599
00:39:26,360 --> 00:39:30,920
What are three things all 
teachers need to know about you 

600
00:39:30,920 --> 00:39:33,120
think? 
Yeah, these are the three sort 

601
00:39:33,120 --> 00:39:34,920
of products. 
Not, maybe not even products, 

602
00:39:34,920 --> 00:39:38,480
but types of product that you 
should be spending your time 

603
00:39:38,480 --> 00:39:41,680
reading about nigh to get ahead 
of the curve or even just to 

604
00:39:41,680 --> 00:39:46,240
catch up. 
OK, I'll do 3, then a bonus one.

605
00:39:47,920 --> 00:39:54,760
So for the teacher, if you were 
to go sort of 1, one tool to 

606
00:39:54,760 --> 00:39:56,880
rule them all sort of thing, one
place to go. 

607
00:39:57,760 --> 00:40:02,600
So there are must be hundreds of
start-ups starting up sort of 

608
00:40:02,600 --> 00:40:07,160
every every day. 
But if I think you know you 

609
00:40:07,160 --> 00:40:11,400
really can't go, it's it feels 
so weird saying this, but you 

610
00:40:11,400 --> 00:40:14,880
really can't go wrong with 
Microsoft Bing as your kind of 

611
00:40:14,880 --> 00:40:19,000
best free at the point of use 
generative AI tool for text. 

612
00:40:19,320 --> 00:40:21,280
And images. 
All we need is a Microsoft 

613
00:40:21,280 --> 00:40:26,000
account and you get you know 
effectively, effectively, you 

614
00:40:26,000 --> 00:40:31,360
know, unlimited conversations 
with this AI chat bot that you 

615
00:40:31,360 --> 00:40:36,400
can use to create lesson plans, 
create quizzes, think of 

616
00:40:36,400 --> 00:40:41,080
explanations, get thinking about
analogies, generates interesting

617
00:40:41,320 --> 00:40:44,000
images that you might want to 
use in lessons. 

618
00:40:44,280 --> 00:40:50,120
For example, the reason why chat
Why Bing is particularly 

619
00:40:50,360 --> 00:40:55,400
effective and useful is because 
Microsoft have plowed a lot of 

620
00:40:55,400 --> 00:41:03,800
money into Open AI and so Bing 
is hooked into the ChatGPT for 

621
00:41:03,800 --> 00:41:07,120
back end. 
So it's using the most up to 

622
00:41:07,120 --> 00:41:14,760
date, certainly the most popular
large language model to generate

623
00:41:15,680 --> 00:41:21,240
text and images because it uses 
the recently updated Dolly 3 

624
00:41:21,240 --> 00:41:23,600
model to help with images as 
well. 

625
00:41:24,720 --> 00:41:30,040
If you usually go through Open 
AI themselves to get access to 

626
00:41:30,040 --> 00:41:33,920
ChatGPT for I think it's 
something like £20 a month I 

627
00:41:33,920 --> 00:41:35,920
think. 
So, you know, you can save 

628
00:41:35,920 --> 00:41:39,040
yourself a lot of money straight
away there by just doing it all 

629
00:41:39,040 --> 00:41:41,840
through all through Bing. 
They're not quite the same. 

630
00:41:41,840 --> 00:41:44,960
They're also going to tweak and 
style it quite differently. 

631
00:41:44,960 --> 00:41:50,920
But you know effectively, and 
this is kind of an important 

632
00:41:50,920 --> 00:41:56,240
point as well, kind of buried 
within this one, is that on any 

633
00:41:56,240 --> 00:42:04,560
AI tool that comes up or that 
you may come across, have a look

634
00:42:04,560 --> 00:42:09,520
and see what it's large language
model it's using to draw its 

635
00:42:09,520 --> 00:42:11,680
data from. 
Because if it's using the same 

636
00:42:11,680 --> 00:42:15,480
back end, you can probably get 
the same answer using a free 

637
00:42:15,480 --> 00:42:19,160
version. 
It may not look as pretty, but 

638
00:42:19,160 --> 00:42:24,600
what you actually ask for the 
content is I would say 9095% 

639
00:42:24,600 --> 00:42:29,040
going to be the same. 
So if you haven't used any tools

640
00:42:29,040 --> 00:42:31,640
yet, you don't know where to 
start or you've perhaps used 

641
00:42:31,640 --> 00:42:37,280
ChatGPT 3.5, the free version, I
would definitely say it's worth 

642
00:42:37,320 --> 00:42:41,880
having a go at looking at 
Microsoft Bing because as I say,

643
00:42:41,880 --> 00:42:48,320
it's free and has the most up to
date models that it draws upon. 

644
00:42:48,320 --> 00:42:53,320
So that's definitely going to be
high up there linked to this. 

645
00:42:53,440 --> 00:42:55,640
So these are all nicely linked 
together. 

646
00:42:56,720 --> 00:43:01,640
The kind of second thing is that
there are two classes of sort of

647
00:43:01,880 --> 00:43:05,120
AI tools sort of out there right
now. 

648
00:43:06,760 --> 00:43:08,840
These are names that I've kind 
of given them. 

649
00:43:09,120 --> 00:43:12,400
They're not ubiquitous by any 
sense, the imagination, but 

650
00:43:12,400 --> 00:43:15,840
there's a sort of a closed AI 
tool. 

651
00:43:16,600 --> 00:43:22,080
What I mean by that is that the 
prompt is effectively invisible 

652
00:43:22,080 --> 00:43:25,440
from you and all you will be 
asked to do is input some 

653
00:43:25,440 --> 00:43:29,040
information. 
So the stuff that Oak had put 

654
00:43:29,040 --> 00:43:34,640
out recently, that's a good 
example of a closed AI tool 

655
00:43:34,640 --> 00:43:40,240
because all you do is ask for 
what year, group, subjects, any 

656
00:43:40,240 --> 00:43:42,960
particular topic that you want 
which you can input yourself, 

657
00:43:43,920 --> 00:43:48,320
you press enter and sort of you 
know there you go. 

658
00:43:49,120 --> 00:43:51,200
It kind of brings you back 
something and there are you know

659
00:43:51,200 --> 00:43:54,720
some advantages to that. 
One is that it's, you know, 

660
00:43:54,840 --> 00:43:57,880
quick and easy. 
The the user doesn't have to 

661
00:43:57,880 --> 00:44:01,200
learn how to do effective 
prompting. 

662
00:44:01,520 --> 00:44:02,880
And as I say, it's quick and 
easy. 

663
00:44:02,880 --> 00:44:06,440
But there are sort of some 
downsides to that in that it 

664
00:44:06,440 --> 00:44:11,480
doesn't sort of offer a serve 
and return type interface. 

665
00:44:11,480 --> 00:44:13,600
So what you're given is what you
get. 

666
00:44:13,600 --> 00:44:16,760
You can't necessarily go back 
and tell the AIO, I don't really

667
00:44:16,760 --> 00:44:19,880
like question 4. 
Can you please change that to 

668
00:44:19,960 --> 00:44:23,680
something else or I really 
wanted you to include the Norman

669
00:44:23,680 --> 00:44:26,680
started building castles quite 
quickly when they invaded in 

670
00:44:26,680 --> 00:44:28,040
1066. 
You've missed that. 

671
00:44:28,040 --> 00:44:30,600
Can you please put that in your 
explanation? 

672
00:44:30,720 --> 00:44:32,800
So you can't kind of do that. 
So it kind of really kind of 

673
00:44:32,800 --> 00:44:35,160
limits you. 
It gets you there quickly, but 

674
00:44:35,160 --> 00:44:38,200
sort of really limits you as to 
what you can do. 

675
00:44:39,440 --> 00:44:44,680
The other type which is the type
of tool that Bing is that 

676
00:44:44,680 --> 00:44:50,680
ChatGPT is that Dolly is is what
I kind of call open tools in 

677
00:44:50,680 --> 00:44:56,200
that you need to then provide 
that prompt so you're in 

678
00:44:56,200 --> 00:44:58,560
complete control as to what that
prompt should be. 

679
00:44:58,560 --> 00:45:02,520
And again some advantages 
disadvantages are is that well 

680
00:45:03,000 --> 00:45:05,160
you're not limited to the type 
of requests. 

681
00:45:05,160 --> 00:45:11,320
So loads of these tools the way 
they sort of they'll usually say

682
00:45:11,320 --> 00:45:13,600
oh you know we have a lesson 
planner tool, a quiz creator 

683
00:45:13,600 --> 00:45:15,680
tool and you can have that for 
free. 

684
00:45:15,680 --> 00:45:18,640
But if you want the assembly 
creator tool and you want the 

685
00:45:19,280 --> 00:45:21,640
school improvement plan tool, 
then you know it's, you know, 

686
00:45:21,640 --> 00:45:25,360
699 a month, you know extra 
whatever it is, if they're using

687
00:45:25,360 --> 00:45:27,960
ChatGPT 4, you know you don't 
need to worry about any of that 

688
00:45:27,960 --> 00:45:30,840
because you can control the 
prompt yourself. 

689
00:45:31,680 --> 00:45:33,840
So you can just ask for those 
things because all of those, 

690
00:45:34,200 --> 00:45:37,640
what those types of applications
and tools are doing are just 

691
00:45:37,640 --> 00:45:39,760
effectively keeping that prompt 
hidden from you. 

692
00:45:39,800 --> 00:45:43,880
If you have good prompting 
engineering, you know you'll be 

693
00:45:43,880 --> 00:45:48,440
able to get, you know, probably 
80% of what they give back to 

694
00:45:48,440 --> 00:45:51,920
you. 
So that's why it's good because 

695
00:45:51,920 --> 00:45:53,480
you can prompt for anything that
you want. 

696
00:45:53,480 --> 00:45:56,640
You're only limited by what you 
yourself can think of. 

697
00:45:57,680 --> 00:46:00,320
You have complete control of 
what that input is. 

698
00:46:00,600 --> 00:46:03,680
A big bug bear of mine is that 
and this goes on to my next 

699
00:46:04,120 --> 00:46:06,960
thing that teachers should know.
But because most of these things

700
00:46:06,960 --> 00:46:10,480
are trained on American data and
sort of American and they have, 

701
00:46:10,480 --> 00:46:13,800
you know, biases built in built 
into them, a lot of the output 

702
00:46:13,800 --> 00:46:18,160
is going to be American English.
So like my favorite, the ending 

703
00:46:18,160 --> 00:46:20,320
of a prompt is always say the 
output must be in British 

704
00:46:20,320 --> 00:46:23,560
English. 
You know, small tiny thing, but 

705
00:46:23,560 --> 00:46:26,280
it's going to, you know, save 
you a bit of time. 

706
00:46:27,360 --> 00:46:30,240
Can't do that with the other 
ones unless behind the scenes 

707
00:46:30,240 --> 00:46:32,400
they've also coded for that. 
It must be in British English, 

708
00:46:32,400 --> 00:46:36,880
but you will never know that 
until you've got a prompt. 

709
00:46:37,320 --> 00:46:41,240
But put your information in and 
got some data routes, it does 

710
00:46:41,240 --> 00:46:42,840
allow for that kind of serve and
return. 

711
00:46:42,840 --> 00:46:45,760
So if I put a prompt in and it 
comes back with something that I

712
00:46:45,760 --> 00:46:48,240
don't like, I can quickly say, 
hey, I really liked this bit, 

713
00:46:48,240 --> 00:46:50,840
this bit, this bit, but I'd 
really appreciate it if you 

714
00:46:50,840 --> 00:46:53,240
included this. 
So you can kind of then sort of,

715
00:46:53,360 --> 00:46:56,320
you know, narrow down what you 
want and then really narrow your

716
00:46:56,320 --> 00:47:00,160
focus. 
It develops for the teacher 

717
00:47:00,160 --> 00:47:02,840
expertise in prompting. 
So they actually understand the 

718
00:47:02,840 --> 00:47:05,680
underlying features and 
structures of what you know and 

719
00:47:05,680 --> 00:47:09,040
their mental models of what this
technology is and how it's going

720
00:47:09,040 --> 00:47:12,400
to work best for them, something
that doesn't necessarily happen 

721
00:47:12,400 --> 00:47:17,240
with these closed tools. 
And let's say going back to what

722
00:47:17,240 --> 00:47:20,680
we said, you know, is this going
to get rid of our thinking, It's

723
00:47:20,680 --> 00:47:23,840
definitely not because the user 
in these open tools can use 

724
00:47:23,840 --> 00:47:27,880
their own expertise to their 
subject phase, the curriculum to

725
00:47:27,880 --> 00:47:31,080
guide these large language 
models to the output that kind 

726
00:47:31,080 --> 00:47:35,520
of best suits their context. 
And so don't necessarily get 

727
00:47:35,520 --> 00:47:39,880
that again with that closed sort
of tools. 

728
00:47:40,400 --> 00:47:43,240
Now obviously there are some 
disadvantage to this one as 

729
00:47:43,240 --> 00:47:45,400
well. 
There is a big learning gap. 

730
00:47:45,400 --> 00:47:49,440
You have to kind of learn how to
prompt and what effective 

731
00:47:49,440 --> 00:47:54,360
prompts are and that takes time.
And so obviously that's then 

732
00:47:54,360 --> 00:47:57,760
going to perhaps impede how 
quickly you get started with 

733
00:47:57,760 --> 00:47:59,360
this. 
If you don't feel successful 

734
00:47:59,360 --> 00:48:02,120
straight away in generating what
you want, you might feel a bit 

735
00:48:02,120 --> 00:48:06,000
disheartened and not motivated 
to give it a go. 

736
00:48:07,400 --> 00:48:14,640
But I think the initial effort 
is worth the outcome by using. 

737
00:48:15,840 --> 00:48:19,960
And again, all biased here. 
I have a newsletter that kind of

738
00:48:19,960 --> 00:48:24,320
tries to teach you how to do 
this open style use of AI tools.

739
00:48:24,320 --> 00:48:29,440
But I think in the long run, you
know, it's better for teachers 

740
00:48:29,960 --> 00:48:33,640
to use that than some of these 
closed tools. 

741
00:48:35,120 --> 00:48:39,320
And the kind of the third thing,
again, we've kind of talked 

742
00:48:39,320 --> 00:48:41,360
about this before, but it's 
really important, is that it's 

743
00:48:41,360 --> 00:48:45,600
not always right. 
So you may think that you've 

744
00:48:45,600 --> 00:48:48,720
delivered something, got 
something back that's, you know,

745
00:48:48,960 --> 00:48:52,280
perfect. 
I think I saw something recently

746
00:48:52,280 --> 00:48:56,920
on on Twitter about someone who 
wanted some descriptions of, I 

747
00:48:57,960 --> 00:49:02,920
think it was like gothic castles
or something like that. 

748
00:49:02,920 --> 00:49:05,280
And she wanted it from from 
Dracula. 

749
00:49:05,280 --> 00:49:10,280
So she wrote, I mean again, a 
prompt of, you know, to try and 

750
00:49:10,280 --> 00:49:14,240
get that information back and 
ChatGPT, whatever she used, 

751
00:49:15,760 --> 00:49:19,640
brought back an excellent 
description of a Gothic castle 

752
00:49:19,640 --> 00:49:24,800
in relative style that I as a as
a novice in Gothic literature 

753
00:49:24,800 --> 00:49:27,160
would not be able to say, well, 
that's definitely not, you know,

754
00:49:27,200 --> 00:49:30,520
Bram Stoker. 
But you know, one of the first 

755
00:49:30,520 --> 00:49:34,200
comments was, hey, you know, 
that's actually like not the 

756
00:49:34,200 --> 00:49:37,400
description that's used in 
Dracula, right. 

757
00:49:37,400 --> 00:49:39,160
And you know, that was a 
surprise to her. 

758
00:49:39,480 --> 00:49:44,080
So it's never right because it's
not connected to the Internet 

759
00:49:44,080 --> 00:49:47,120
necessarily. 
It is just a prediction device. 

760
00:49:47,120 --> 00:49:52,920
It will be really, you know, 
it's well, they always say like 

761
00:49:52,920 --> 00:49:54,960
as long as you say anything with
enough confidence, like you'll 

762
00:49:55,000 --> 00:49:56,440
they'll anyone will believe 
anything. 

763
00:49:56,440 --> 00:50:00,520
Like it is literally like that 
Incarnate. 

764
00:50:00,640 --> 00:50:04,200
It's so good at making you think
that what it's producing is the 

765
00:50:04,200 --> 00:50:07,320
right thing and it's trained on 
data. 

766
00:50:08,160 --> 00:50:17,000
Human created texts and humans 
have biases that will have 

767
00:50:17,000 --> 00:50:19,840
implications for the output of 
that data. 

768
00:50:19,840 --> 00:50:24,000
And so you really kind of have 
to check and make sure that one,

769
00:50:24,000 --> 00:50:28,560
it's accurate. 
And second of all, that humans, 

770
00:50:28,560 --> 00:50:35,520
natural cognitive biases aren't 
necessarily drawn out of that 

771
00:50:35,520 --> 00:50:40,000
output as well because you could
get yourself in a bit of trouble

772
00:50:40,000 --> 00:50:42,320
I think if you don't check for 
that one as well. 

773
00:50:42,320 --> 00:50:47,760
And you do end up producing text
that's may provide a certain 

774
00:50:47,760 --> 00:50:52,040
interpretation that you know you
believe something that you or 

775
00:50:52,040 --> 00:50:56,080
you've provided some data or 
some information you know to 

776
00:50:56,160 --> 00:50:59,080
pupils that isn't necessarily 
correct because it is. 

777
00:50:59,520 --> 00:51:03,520
You know the data was biased, so
it's produced a biased output. 

778
00:51:04,360 --> 00:51:07,760
And the kind of the bonus one 
obviously is that teachers 

779
00:51:07,760 --> 00:51:11,400
should definitely sign up to 
Teacher Prompts newsletter, 

780
00:51:11,920 --> 00:51:14,800
because that will definitely 
tell you everything that you 

781
00:51:14,800 --> 00:51:16,760
need to know. 
I'm currently working on a 

782
00:51:16,840 --> 00:51:24,120
beginner's guide to AI and like 
PEPS did with evidence, Snacks, 

783
00:51:24,440 --> 00:51:26,560
you know, if you refer. 
I think it's going to literally 

784
00:51:26,560 --> 00:51:30,200
be like if you prefer two people
to this, you know you'll get the

785
00:51:30,200 --> 00:51:32,080
guide. 
If you refer six people you'll 

786
00:51:32,080 --> 00:51:37,080
get the the basic image guide as
well. 

787
00:51:37,080 --> 00:51:40,120
So that's that's my half term 
job, so I'm hoping that will be 

788
00:51:40,120 --> 00:51:43,640
done by definitely not by 
tomorrow but hopefully next week

789
00:51:44,160 --> 00:51:47,280
Once everyone's given at the OK 
that will be available there as 

790
00:51:47,280 --> 00:51:49,520
well for people to use. 
I love the. 

791
00:51:49,520 --> 00:51:53,400
I love the. 
Yeah, the incentivized sort of 

792
00:51:53,400 --> 00:51:53,800
thing. 
Yeah. 

793
00:51:53,800 --> 00:51:56,800
And I remember speaking about 
sort of just watching him with 

794
00:51:56,800 --> 00:52:00,200
his chapter from his upcoming 
book and stuff like that there. 

795
00:52:00,200 --> 00:52:02,880
So that I think that's a really 
interesting way to. 

796
00:52:03,640 --> 00:52:05,520
Encourage engagement. 
And I'm looking forward. 

797
00:52:05,520 --> 00:52:08,640
I mean, I'm excited about the 
potential for artificial 

798
00:52:08,640 --> 00:52:10,640
intelligence. 
Just off the back of this, what 

799
00:52:11,200 --> 00:52:14,440
about an Irish conversation? 
So, you know, whenever we finish

800
00:52:14,440 --> 00:52:16,600
up, I'm going away and I'm going
to try and read as much about 

801
00:52:16,600 --> 00:52:20,240
engineer prompt engineering as I
possibly can. 

802
00:52:20,240 --> 00:52:21,400
You know what? 
What do you do for a job? 

803
00:52:21,400 --> 00:52:25,480
Well, I'm a prompt engineer, you
know, they say it's going to be 

804
00:52:25,480 --> 00:52:31,520
a very highly valuable 
profession, since I think that 

805
00:52:31,520 --> 00:52:35,400
is probably, you know, the jobs 
of tomorrow kind of brigade 

806
00:52:35,400 --> 00:52:39,680
going a little bit too far, but 
it's definitely an interesting, 

807
00:52:40,080 --> 00:52:42,400
interesting Ave. 
Yeah. 

808
00:52:42,400 --> 00:52:44,800
I mean, if there's one thing 
I've learned since starting my 

809
00:52:44,800 --> 00:52:48,000
own business recently is that if
there's anything that could make

810
00:52:48,360 --> 00:52:51,600
your life easier, someone's got 
that corner of the market sewn 

811
00:52:51,600 --> 00:52:54,240
up already. 
So if you want to save yourself 

812
00:52:54,240 --> 00:52:58,040
some some money. 
And get exactly what you need. 

813
00:52:58,080 --> 00:53:00,240
And then I think the prompt 
engineer's going to be really, 

814
00:53:00,240 --> 00:53:04,280
really important for getting the
most of them from from 

815
00:53:04,280 --> 00:53:09,800
artificial intelligence. 
What's your bold prediction? 

816
00:53:09,840 --> 00:53:14,160
For the next 12 months, the 
outputs from these models are 

817
00:53:14,160 --> 00:53:20,600
only going to be as good as the 
data that the model uses to 

818
00:53:20,600 --> 00:53:28,040
learn from and Oak National, 
because all of their resources 

819
00:53:28,040 --> 00:53:31,080
are open source. 
Effectively, large language 

820
00:53:31,080 --> 00:53:37,320
models can comb everything that 
Oak has will create. 

821
00:53:38,520 --> 00:53:43,280
And so that means that there is 
going to be, you know, models 

822
00:53:43,960 --> 00:53:48,520
that are trained specifically on
education data and specifically 

823
00:53:48,520 --> 00:53:53,560
education data of the the 2014 
National Curriculum for England.

824
00:53:53,560 --> 00:53:57,960
And so as working in education 
in England, if you have models 

825
00:53:57,960 --> 00:54:00,960
that are trained specifically on
that data, that's going to be, 

826
00:54:01,000 --> 00:54:05,480
you know, particularly useful 
for teachers and that's where 

827
00:54:05,520 --> 00:54:08,880
whatever Oak do next could be 
really quite interesting over 

828
00:54:08,880 --> 00:54:12,640
the next 12 months. 
So I think bold prediction there

829
00:54:12,640 --> 00:54:16,360
is to kind of keep an eye out on
what Oak are doing because 

830
00:54:17,240 --> 00:54:19,080
they're going to produce 
something I think that could be 

831
00:54:19,080 --> 00:54:22,040
really useful and valuable to 
teachers. 

832
00:54:22,040 --> 00:54:24,200
I hope they open it up. 
So it's not just a kind of a 

833
00:54:24,200 --> 00:54:29,840
closed tool, but having it 
specifically trained on, having 

834
00:54:29,840 --> 00:54:32,040
it specifically trained on 
education today that's 

835
00:54:32,160 --> 00:54:34,920
completely relevant to, you 
know, teachers in England is 

836
00:54:34,920 --> 00:54:37,440
going to be quite a powerful 
thing, but they're all going to 

837
00:54:37,440 --> 00:54:39,880
open that data up to other 
models as well. 

838
00:54:39,880 --> 00:54:43,360
So you know, it may not just be 
Oak that has access to all of 

839
00:54:43,360 --> 00:54:45,480
that. 
The biggest weakness that I've 

840
00:54:45,480 --> 00:54:50,400
found in generative image 
creation right now is that it is

841
00:54:50,400 --> 00:54:53,480
spectacularly poor educational 
diagrams. 

842
00:54:53,880 --> 00:54:59,720
So if I want were to ask it to 
create a show me a bar model of 

843
00:55:01,200 --> 00:55:05,960
4 * 6 or, you know, here's the 
here's a word problem, show me a

844
00:55:05,960 --> 00:55:08,240
bar model that goes with that. 
Can't do it. 

845
00:55:10,520 --> 00:55:13,240
So I'm kind of confident in the 
next 12 months. 

846
00:55:13,640 --> 00:55:16,080
Well, I feel not confident. 
I don't have no insider 

847
00:55:16,080 --> 00:55:18,600
knowledge, but I feel like it's 
an easy win for them to like if 

848
00:55:18,600 --> 00:55:21,480
you nail the educational 
diagram, you're going to make a 

849
00:55:21,480 --> 00:55:25,680
lot of teachers very happy. 
If I can just say, you know, 

850
00:55:26,040 --> 00:55:28,760
consider the split attention 
effect and create a simple 

851
00:55:28,760 --> 00:55:33,160
diagram of the water cycle aimed
for pupils in, you know, year 

852
00:55:33,160 --> 00:55:38,000
four who are in the, you know, 
studying the English curriculum.

853
00:55:38,560 --> 00:55:41,320
And I might say, you know, it 
must include, you know, 

854
00:55:41,360 --> 00:55:44,560
evaporation, condensation 
runoff, whatever else you want 

855
00:55:44,560 --> 00:55:46,320
it to include. 
Enter. 

856
00:55:46,320 --> 00:55:49,800
And then it produces a useful 
graphic that takes into account 

857
00:55:49,800 --> 00:55:52,560
the split attention effect. 
So it's not key in the corner. 

858
00:55:52,560 --> 00:55:55,280
It's, you know, labeled where it
needs to be. 

859
00:55:55,280 --> 00:55:57,960
And I don't have to worry about 
the copyright of that. 

860
00:55:57,960 --> 00:56:00,000
That's going to be really 
powerful for teachers. 

861
00:56:01,120 --> 00:56:03,960
Same way for, you know, you and 
your math journey. 

862
00:56:03,960 --> 00:56:06,040
I'm sure if you could say, you 
know, I want you to show them 

863
00:56:06,440 --> 00:56:10,760
this model using algebra tiles 
or you know, cuisinare rods, 

864
00:56:10,760 --> 00:56:13,680
whatever it might be and you 
could get generative AI to do 

865
00:56:13,680 --> 00:56:16,240
that, you know, that'd be really
powerful. 

866
00:56:16,240 --> 00:56:18,520
I've told Johnny, you know, he 
needs to get in that market 

867
00:56:18,520 --> 00:56:23,240
really, really quickly, make it 
happen because no one's doing it

868
00:56:23,240 --> 00:56:28,920
right now. 
And kind of the big the next 

869
00:56:28,920 --> 00:56:34,440
kind of prediction is that I do 
kind of think you know I think 

870
00:56:34,440 --> 00:56:40,680
more teachers are going to just 
keep on using it and find 

871
00:56:40,880 --> 00:56:46,120
interesting ways to you know 
make it work and make it make it

872
00:56:46,120 --> 00:56:50,040
work for them make it useful and
hopefully cuts down a little bit

873
00:56:50,040 --> 00:56:51,800
on their on their teacher 
workload. 

874
00:56:51,800 --> 00:56:56,800
And if I can play but a but a 
small part in that for what is 

875
00:56:56,800 --> 00:57:01,920
for me, you know hours work on a
weekend to put that newsletter 

876
00:57:01,920 --> 00:57:06,400
together, Be a happy man, yeah. 
Because at the minute, obviously

877
00:57:06,400 --> 00:57:09,480
we made those movie posters. 
We know that it currently 

878
00:57:09,480 --> 00:57:14,000
struggles with thin words. 
So there are some hurdles to 

879
00:57:14,000 --> 00:57:19,760
overcome before diagrams and 
models are a realistic prospect.

880
00:57:19,960 --> 00:57:23,440
But yeah, but imagine being able
to sort of put in the prompt and

881
00:57:23,440 --> 00:57:27,000
then you would have that 
representation ready for you. 

882
00:57:27,000 --> 00:57:29,160
I think it would open up a lot 
of doors. 

883
00:57:29,320 --> 00:57:31,120
So fingers crossed. 
I mean obviously. 

884
00:57:32,440 --> 00:57:35,080
No more marking or sort of 
considering how they can get the

885
00:57:35,080 --> 00:57:36,760
most from it. 
So it'd be interesting to see 

886
00:57:36,760 --> 00:57:38,880
what happens with them because 
they've got a massive data set 

887
00:57:39,520 --> 00:57:44,360
on which to train and stuff. 
I don't know your your reaction 

888
00:57:44,360 --> 00:57:47,240
seems to think that they haven't
started exploring this and have 

889
00:57:47,240 --> 00:57:49,920
we just discussed that they 
should explore this or I 

890
00:57:49,920 --> 00:57:51,560
thought, no they're doing 
they're doing some interesting 

891
00:57:51,560 --> 00:57:53,000
stuff. 
No more marking. 

892
00:57:53,320 --> 00:57:53,960
Absolutely. 
Yeah. 

893
00:57:55,440 --> 00:58:00,160
Daisy is a is an intellectual 
powerhouse and she's very good 

894
00:58:00,160 --> 00:58:05,600
at getting rid of the the rub, 
finding the bits of gold within 

895
00:58:05,600 --> 00:58:08,800
all the all the nonsense that 
are out that's out there right 

896
00:58:08,800 --> 00:58:10,680
now. 
And they do some, Yeah, they do.

897
00:58:10,760 --> 00:58:11,920
They do good stuff. 
Cool. 

898
00:58:11,920 --> 00:58:14,440
Well, like I say, it's an 
absolute pleasure and I think 

899
00:58:14,440 --> 00:58:18,400
I've learned a lot. 
I am excited about the next 12 

900
00:58:18,400 --> 00:58:20,320
months, really looking forward 
to seeing what happens and 

901
00:58:20,360 --> 00:58:23,760
definitely going to read more. 
I'm going to sort of pick your 

902
00:58:23,760 --> 00:58:26,880
brainer offline for where you 
going for your information first

903
00:58:26,880 --> 00:58:29,200
time because the minute I'm 
getting it from from guys here 

904
00:58:29,200 --> 00:58:31,720
financially. 
Incentivized for this to be a 

905
00:58:31,720 --> 00:58:35,640
success, You know, so they're 
all very much in Palo Alto and 

906
00:58:35,640 --> 00:58:40,080
and the surrounding suburbs and 
and have a, you know, have a 

907
00:58:40,120 --> 00:58:42,320
good skin in the game. 
You know, they're not in 

908
00:58:42,320 --> 00:58:45,800
education fields, but you know, 
they're incentivized to make 

909
00:58:45,800 --> 00:58:48,360
sure that technology continues 
to develop and stuff. 

910
00:58:48,360 --> 00:58:50,520
And so it'd be interesting to 
find out, you know, because I 

911
00:58:50,520 --> 00:58:55,080
doubt research papers exist, you
know, because it's happening in 

912
00:58:55,080 --> 00:58:57,800
real time, so to speak. 
You know be surprised if you go 

913
00:58:57,800 --> 00:59:00,600
on to if you go on to our our 
litmaps. 

914
00:59:00,600 --> 00:59:04,840
There's a there's an area 
there's a thing that I check 

915
00:59:04,840 --> 00:59:07,800
every now and then about ChatGPT
and engineer prompting and stuff

916
00:59:07,800 --> 00:59:10,120
like that that I kind of put 
together. 

917
00:59:10,400 --> 00:59:19,240
Ethan, I think it's Malik. 
He's he's a pretty sensible AI 

918
00:59:20,040 --> 00:59:22,880
engineer prompter researcher 
chap. 

919
00:59:22,880 --> 00:59:25,040
He's worth. 
He's worth a follow on Twitter 

920
00:59:25,040 --> 00:59:26,760
if you don't. 
He's got a sub stack too, which 

921
00:59:26,760 --> 00:59:29,320
is worth a with a subscription 
too. 

922
00:59:29,680 --> 00:59:34,560
And he's a tech guru. 
No, he's a teacher. 

923
00:59:35,080 --> 00:59:39,680
He's a lecturer at university in
America. 

924
00:59:40,520 --> 00:59:43,200
So he's not in the tech. 
He's a he's a one of us, not, 

925
00:59:43,960 --> 00:59:48,120
not a tech guy one. 
Of US Independent Adjudicator, 

926
00:59:48,880 --> 00:59:50,520
Independent Adjudicator, 
absolutely. 

927
00:59:51,040 --> 00:59:54,640
Nice, excellent. 
Well, I mean to be continued I 

928
00:59:54,640 --> 00:59:57,040
think, yeah, see another 12 
months. 

929
00:59:59,280 --> 01:00:02,320
Well, it could be some sort of 
robot hosting the show by then. 

930
01:00:02,720 --> 01:00:05,200
Well, fingers crossed, fingers 
crossed. 

931
01:00:05,600 --> 01:00:08,720
Yeah, that's the other, Yeah, we
haven't explored the the 

932
01:00:08,720 --> 01:00:12,040
vocalization stuff yet, have we?
No, that's for another time, 

933
01:00:12,040 --> 01:00:15,520
because I want to see how we can
work that in this episode never 

934
01:00:15,520 --> 01:00:17,960
really happened. 
It was all just an AI voice. 

935
01:00:19,720 --> 01:00:21,400
I mean, it's it struggles with 
my voice. 

936
01:00:21,600 --> 01:00:23,520
It's got your itch. 
It's your voice down pat. 

937
01:00:24,320 --> 01:00:25,720
Yeah, it does it pretty well, 
doesn't it? 

938
01:00:26,200 --> 01:00:27,920
Yeah, but it thinks I'm 
American. 

939
01:00:28,680 --> 01:00:31,840
Well, welcome to thinking deeply
about primary education. 

940
01:00:32,480 --> 01:00:33,640
That doesn't sound like me at 
all. 

941
01:00:34,360 --> 01:00:35,600
No, that doesn't. 
To be fair. 

942
01:00:36,600 --> 01:00:38,480
All the set to do then is Seth. 
Thank you very much for joining 

943
01:00:38,480 --> 01:00:40,280
me. 
Thank you so much for having me,

944
01:00:40,280 --> 01:00:41,280
Karen. 
Appreciate it. 

945
01:00:41,800 --> 01:00:43,960
And everyone at home. 
Until next time, thanks for 

946
01:00:43,960 --> 01:00:44,360
listening.
