WEBVTT
Language: en-US

1
00:00:01.649 --> 00:00:06.369
<v Speaker 1>Welcome to the trend detection podcast,
powered by Siemens, and now your host, Liz

2
00:00:06.369 --> 00:00:06.769
<v Speaker 1>McGinn.

3
00:00:09.010 --> 00:00:09.750
<v Liz McGinn>Hello, everyone.

4
00:00:09.810 --> 00:00:11.810
<v Liz McGinn>This is Liz McGinn with Trend Detection
podcast.

5
00:00:11.880 --> 00:00:13.250
<v Liz McGinn>I'm here joined by James.

6
00:00:13.369 --> 00:00:16.969
<v Liz McGinn>We are going to talk about machine
learning and AI and what's the difference, and he

7
00:00:16.969 --> 00:00:17.829
<v Liz McGinn>is the teacher today.

8
00:00:18.010 --> 00:00:21.129
<v Liz McGinn>So, James, you want to kind of introduce
yourself and why you're qualified to speak

9
00:00:21.129 --> 00:00:21.729
<v Liz McGinn>on the topic?

10
00:00:23.350 --> 00:00:23.760
<v James Loach>OK.

11
00:00:24.040 --> 00:00:24.239
<v James Loach>OK.

12
00:00:24.319 --> 00:00:24.799
<v James Loach>Hi, Liz.

13
00:00:25.069 --> 00:00:26.899
<v James Loach>Yeah, so, um, I'm James Loach.

14
00:00:27.000 --> 00:00:32.459
<v James Loach>I'm head of research at Sensai and have
been since the early days of the company,

15
00:00:32.880 --> 00:00:36.540
<v James Loach>um, and I'm also, um, part of the, the
product leadership team.

16
00:00:36.840 --> 00:00:41.549
<v James Loach>You know, as to why I'm qualified to talk
on, on this topic, I think this would be,

17
00:00:41.560 --> 00:00:45.319
<v James Loach>um, something down to practical experience
since I've been building machine learning

18
00:00:45.319 --> 00:00:47.779
<v James Loach>AI systems, you know, for, for, for that
period.

19
00:00:48.180 --> 00:00:50.860
<v James Loach>Uh, my background before that is in
something different though, which is

20
00:00:50.860 --> 00:00:52.689
<v James Loach>experimental particle physics.

21
00:00:52.909 --> 00:00:56.470
<v James Loach>Um, but in terms of the way you think
about the world and the actual things that

22
00:00:56.470 --> 00:01:00.790
<v James Loach>you're doing, it's not necessarily too,
too different from what I'm doing now in the

23
00:01:00.790 --> 00:01:02.150
<v James Loach>machine learning and AI, you know.

24
00:01:03.209 --> 00:01:03.590
<v Liz McGinn>Nice.

25
00:01:03.849 --> 00:01:04.809
<v Liz McGinn>That's interesting.

26
00:01:05.559 --> 00:01:10.000
<v Liz McGinn>Um, I guess to jump into it, when we're
talking about artificial intelligence and

27
00:01:10.000 --> 00:01:13.160
<v Liz McGinn>machine learning, can you give some
definitions just so everyone's on the same page

28
00:01:13.160 --> 00:01:13.900
<v Liz McGinn>when we first start?

29
00:01:15.669 --> 00:01:19.860
<v James Loach>Yeah, I think you can think of artificial
intelligence as a thing that you're,

30
00:01:19.949 --> 00:01:21.150
<v James Loach>you're aiming for, right?

31
00:01:21.389 --> 00:01:25.690
<v James Loach>So we all know what intelligence is, you
know, for our day to day interactions with

32
00:01:25.690 --> 00:01:29.059
<v James Loach>intelligent beings or animals or all this
kind of stuff, right?

33
00:01:29.389 --> 00:01:31.309
<v James Loach>So this is the kind of thing that that you
want.

34
00:01:31.629 --> 00:01:36.830
<v James Loach>So a system which is able to take in
sensory information from the world and based on

35
00:01:36.830 --> 00:01:41.480
<v James Loach>that information take intelligent actions
in the world and, you know, continue on

36
00:01:41.480 --> 00:01:42.069
<v James Loach>like that.

37
00:01:42.699 --> 00:01:46.980
<v James Loach>And machine learning is one way to, um,
achieve that.

38
00:01:47.260 --> 00:01:47.569
<v James Loach>Yeah.

39
00:01:47.709 --> 00:01:53.139
<v James Loach>And machine learning, um, it's about
having systems, um, that, yeah, learn, learn

40
00:01:53.139 --> 00:01:55.379
<v James Loach>from data they take in from the world,
right?

41
00:01:55.500 --> 00:01:58.879
<v James Loach>So it's, it's a model that you're training
on data from the world and this model

42
00:01:58.989 --> 00:02:01.099
<v James Loach>will make, make useful predictions.

43
00:02:01.139 --> 00:02:05.300
<v James Loach>And, in the case of something like an
animal, um, in a natural system, these will be

44
00:02:05.300 --> 00:02:06.239
<v James Loach>the actions to take.

45
00:02:06.790 --> 00:02:10.750
<v James Loach>Um, and in artificial systems, you know,
the output of the machine learning system

46
00:02:10.750 --> 00:02:15.750
<v James Loach>could be, um, a forecast, it could be a
suggestion of, um, what, what the problem is

47
00:02:15.750 --> 00:02:17.979
<v James Loach>with your machine or in something like
chat GPT.

48
00:02:18.029 --> 00:02:20.570
<v James Loach>It could be the, the message from the, the
helpful assistant.

49
00:02:21.270 --> 00:02:26.070
<v James Loach>OK, so AI is the goal, and machine
learning is one technique that, that we use to,

50
00:02:26.190 --> 00:02:27.649
<v James Loach>to try and realize that goal.

51
00:02:28.139 --> 00:02:33.529
<v Liz McGinn>OK, so machine learning is used to build
artificial intelligence.

52
00:02:33.910 --> 00:02:35.850
<v Liz McGinn>Yes, yeah, awesome.

53
00:02:36.610 --> 00:02:40.100
<v Liz McGinn>Um, I guess That kind of skips my next
question.

54
00:02:40.389 --> 00:02:45.669
<v Liz McGinn>Um, do you, so I noticed that a lot of
these terms are used interchangeably, um,

55
00:02:45.750 --> 00:02:47.750
<v Liz McGinn>with artificial intelligence and machine
learning.

56
00:02:47.830 --> 00:02:51.979
<v Liz McGinn>Um, is that, um, correct in the
assumption?

57
00:02:52.029 --> 00:02:55.149
<v Liz McGinn>I know, like, technically it's building
up, but like, when we're talking about it in

58
00:02:55.149 --> 00:02:59.550
<v Liz McGinn>the media or within, you know, products or
anything like that, um.

59
00:03:00.470 --> 00:03:02.070
<v Liz McGinn>How do you go about it?

60
00:03:03.169 --> 00:03:08.669
<v James Loach>Yeah, I think at this, this point in
history, the terms are almost synonymous,

61
00:03:08.869 --> 00:03:09.179
<v James Loach>right?

62
00:03:09.429 --> 00:03:13.979
<v James Loach>So they are formally different things, but
you do not come across many systems in

63
00:03:13.979 --> 00:03:18.470
<v James Loach>the world now either claiming to be AI or
trying to achieve that that are not driven

64
00:03:18.470 --> 00:03:19.869
<v James Loach>by machine learning.

65
00:03:20.229 --> 00:03:20.490
<v James Loach>OK.

66
00:03:20.669 --> 00:03:22.199
<v James Loach>This is a little bit different
historically.

67
00:03:22.550 --> 00:03:26.179
<v James Loach>So if you go back to the 60s and 70s, you
know, particularly, there were, there were

68
00:03:26.179 --> 00:03:27.550
<v James Loach>other, other approaches.

69
00:03:27.830 --> 00:03:29.789
<v James Loach>Yeah, um, you know.

70
00:03:30.169 --> 00:03:34.130
<v James Loach>The, the traditional other kind of
approach is where you, you're sort of programming

71
00:03:34.130 --> 00:03:35.570
<v James Loach>in the rules, right?

72
00:03:35.929 --> 00:03:40.289
<v James Loach>So you can imagine such a system, you
know, if the user says this, you process it in

73
00:03:40.289 --> 00:03:45.050
<v James Loach>some way and then you give an answer and
you, you have like complicated rules or

74
00:03:45.050 --> 00:03:46.490
<v James Loach>scripts or things like that.

75
00:03:46.820 --> 00:03:49.940
<v James Loach>And yeah, the thing you do with machine
learning is to get away from all of that, so

76
00:03:49.940 --> 00:03:51.339
<v James Loach>you don't program that stuff in.

77
00:03:51.500 --> 00:03:55.729
<v James Loach>You have the system learn what the
appropriate responses would be, yeah, but these

78
00:03:55.729 --> 00:04:01.589
<v James Loach>days we can make a subtle distinction
between AI and ML, but you know it's not like

79
00:04:01.589 --> 00:04:05.830
<v James Loach>a huge practical importance for the
reasons I described,

80
00:04:05.839 --> 00:04:06.000
<v Liz McGinn>yeah.

81
00:04:07.330 --> 00:04:07.580
<v Liz McGinn>Nice.

82
00:04:07.740 --> 00:04:12.220
<v Liz McGinn>Can you give a little bit more of the
history and the journey, how we got to AI from

83
00:04:12.220 --> 00:04:12.740
<v Liz McGinn>machine learning?

84
00:04:16.510 --> 00:04:21.709
<v James Loach>Um, OK, well, I mean, the, the, the
history of AI is, is difficult to sum up very

85
00:04:21.709 --> 00:04:26.309
<v James Loach>briefly, but, you know, I think the
important things were I can pull out free from,

86
00:04:26.359 --> 00:04:31.640
<v James Loach>from what I just mentioned, right, that
people are starting out with, um, You know,

87
00:04:32.160 --> 00:04:35.880
<v James Loach>machines which, um, work in a symbolic
way, right?

88
00:04:36.040 --> 00:04:41.079
<v James Loach>So, you know, you have different concepts
and operations on those concepts and you

89
00:04:41.079 --> 00:04:46.040
<v James Loach>kind of, you know, get, use, use
information from the world to like, to program all

90
00:04:46.040 --> 00:04:47.130
<v James Loach>this up, right?

91
00:04:47.679 --> 00:04:50.880
<v James Loach>You know, so you can describe these
objects in the world and they have kind of rules

92
00:04:50.880 --> 00:04:53.570
<v James Loach>about what they are and how they interact
and things like that.

93
00:04:54.000 --> 00:04:55.399
<v James Loach>Um, and you can build up.

94
00:04:56.739 --> 00:05:00.510
<v James Loach>Yeah, systems which feel like they've got
a certain amount of intelligence in a sort

95
00:05:00.510 --> 00:05:02.190
<v James Loach>of deterministic symbolic way.

96
00:05:02.269 --> 00:05:06.579
<v James Loach>And the thing that I, I, I, I mentioned in
my previous answer was, um, yeah, famous

97
00:05:06.579 --> 00:05:12.640
<v James Loach>system, I think from the 1960s called
ELISA, yeah, which is like.

98
00:05:13.790 --> 00:05:17.869
<v James Loach>Like a therapist, you know, that you can,
it will ask you how you are and you can

99
00:05:17.869 --> 00:05:21.239
<v James Loach>write something and it can infer from that
like, oh, tell me more about that, uh,

100
00:05:21.429 --> 00:05:24.549
<v James Loach>tell me about your family, or what do you
think about that, etc.

101
00:05:24.869 --> 00:05:27.589
<v James Loach>And at the time, people could be sort of
convinced that there was, there was some

102
00:05:27.589 --> 00:05:30.470
<v James Loach>intelligence behind this, but all
rule-based, you know.

103
00:05:31.059 --> 00:05:35.079
<v James Loach>Um, and what we've seen from that time to
now is this other approach.

104
00:05:36.380 --> 00:05:41.579
<v James Loach>You have to start off as something which
is conceptual, something which works in toy

105
00:05:41.579 --> 00:05:47.700
<v James Loach>situations, something which has had
various problems in making it work at a larger

106
00:05:47.700 --> 00:05:49.820
<v James Loach>scale solved, you know.

107
00:05:50.160 --> 00:05:53.739
<v James Loach>And you have, you have this sort of
theoretical progress over a number of decades

108
00:05:53.880 --> 00:05:58.320
<v James Loach>and at the same time, you know, very
interesting changes in, uh, com, um, hardware,

109
00:05:58.440 --> 00:06:01.799
<v James Loach>right, that, that make it convenient and,
and possible and possible.

110
00:06:02.000 --> 00:06:03.119
<v James Loach>There's a whole story there, right?

111
00:06:03.480 --> 00:06:07.959
<v James Loach>But, you know, you, you have this, this
movement from early success with symbolic

112
00:06:07.959 --> 00:06:12.160
<v James Loach>rules-based systems to the sort of
absolute dominance now of, of machine learning

113
00:06:12.160 --> 00:06:13.040
<v James Loach>things, um.

114
00:06:13.940 --> 00:06:14.429
<v Liz McGinn>Yeah.

115
00:06:14.950 --> 00:06:15.070
<v Liz McGinn>Nice.

116
00:06:15.190 --> 00:06:15.910
<v Liz McGinn>So very input.

117
00:06:16.070 --> 00:06:21.190
<v Liz McGinn>If it says this, then you respond with
this rule-based versus now it'll almost think

118
00:06:21.190 --> 00:06:21.750
<v Liz McGinn>on its own.

119
00:06:23.170 --> 00:06:27.730
<v James Loach>Um, yeah, well, it's using, it has, it
has, in a weird sense rules inside it.

120
00:06:27.799 --> 00:06:32.649
<v James Loach>I mean, like an AI system is operating in
a deterministic way, you know, it's this

121
00:06:32.649 --> 00:06:36.290
<v James Loach>thing with all of these weights and
connections and you push your question through

122
00:06:36.290 --> 00:06:38.970
<v James Loach>it in simple terms and you get your, your
answer out.

123
00:06:39.290 --> 00:06:40.890
<v James Loach>So there's something deterministic in
that.

124
00:06:41.250 --> 00:06:46.109
<v James Loach>But the key point is that human beings
didn't write it, you know, and there's a

125
00:06:46.109 --> 00:06:51.630
<v James Loach>whole industry now of trying to understand
what is going on inside, you know,

126
00:06:51.750 --> 00:06:53.109
<v James Loach>cutting edge AI models.

127
00:06:53.420 --> 00:06:56.070
<v James Loach>People have to study them in the way that
you might study your brain or something

128
00:06:56.070 --> 00:06:58.709
<v James Loach>like that and try and figure out how it's
actually doing what it is doing.

129
00:06:59.470 --> 00:07:00.109
<v James Loach>Um, yeah,

130
00:07:00.390 --> 00:07:00.730
<v Liz McGinn>anyway.

131
00:07:01.230 --> 00:07:02.250
<v Liz McGinn>Interesting.

132
00:07:03.059 --> 00:07:07.910
<v Liz McGinn>That's interesting how we went from, um,
kind of that role base to now we're trying

133
00:07:07.910 --> 00:07:09.809
<v Liz McGinn>to figure out how it's thinking on its
own.

134
00:07:10.899 --> 00:07:15.579
<v James Loach>Yeah And, and, and this is quite
difficult, yeah, but, but it's, it's somehow easier

135
00:07:15.579 --> 00:07:19.019
<v James Loach>than working at how brains work because of
course you can, um, you can do brain

136
00:07:19.019 --> 00:07:23.179
<v James Loach>surgery on AI models very easily, yeah,
and mess around with things and, and see

137
00:07:23.179 --> 00:07:23.880
<v James Loach>what happens.

138
00:07:25.859 --> 00:07:26.260
<v Liz McGinn>That makes sense.

139
00:07:26.320 --> 00:07:29.579
<v Liz McGinn>I mean, brains are very complicated and
it's hard to study your own thing when

140
00:07:29.579 --> 00:07:30.920
<v Liz McGinn>you're a brain yourself.

141
00:07:34.220 --> 00:07:34.660
<v Liz McGinn>Awesome.

142
00:07:34.779 --> 00:07:38.500
<v Liz McGinn>So when we need to translate that to
predictive maintenance, what does that look

143
00:07:38.500 --> 00:07:38.940
<v Liz McGinn>like?

144
00:07:41.890 --> 00:07:44.070
<v James Loach>Um, OK, so.

145
00:07:45.529 --> 00:07:49.890
<v James Loach>Yeah, in, in, in, in, in broad terms, you
know, you've, you've, you've got some

146
00:07:49.890 --> 00:07:51.209
<v James Loach>machines, right?

147
00:07:51.799 --> 00:07:56.200
<v James Loach>Um, they have some history themselves, you
know, how, how they were constructed, you

148
00:07:56.200 --> 00:07:58.920
<v James Loach>know, and that they're doing a particular
task.

149
00:07:59.200 --> 00:08:02.600
<v James Loach>They have some particular history in your
factory, you know, because people have

150
00:08:02.600 --> 00:08:06.359
<v James Loach>done repairs on them, you know, they've
been used in particular ways over periods

151
00:08:06.359 --> 00:08:07.279
<v James Loach>of, of time.

152
00:08:07.600 --> 00:08:07.799
<v James Loach>Yeah.

153
00:08:08.010 --> 00:08:11.399
<v James Loach>So you have all this, this historical
stuff to do with your machines, and then

154
00:08:11.399 --> 00:08:14.950
<v James Loach>you've got a particular thing at the
moment, right, which is something is happening

155
00:08:14.950 --> 00:08:15.579
<v James Loach>on your machine.

156
00:08:16.070 --> 00:08:19.429
<v James Loach>And that could be something you're
observing or it's something in, in the data that

157
00:08:19.429 --> 00:08:20.369
<v James Loach>you're recording.

158
00:08:20.640 --> 00:08:24.420
<v James Loach>But, you know, you've got this whole,
whole set of information and what, what you're

159
00:08:24.420 --> 00:08:29.350
<v James Loach>interested there is, um, is there anything
happening here that I should be worried

160
00:08:29.350 --> 00:08:29.790
<v James Loach>about?

161
00:08:30.230 --> 00:08:30.510
<v James Loach>OK?

162
00:08:30.589 --> 00:08:34.989
<v James Loach>And if you, that's true or you just are
worried about it, what, what should I do?

163
00:08:35.510 --> 00:08:35.760
<v James Loach>OK.

164
00:08:35.890 --> 00:08:39.270
<v James Loach>So you have this whole load of source
information basically, and what you want at

165
00:08:39.270 --> 00:08:43.349
<v James Loach>the end is something like a, a statement
of whether this deserves your attention and

166
00:08:43.349 --> 00:08:45.349
<v James Loach>if it does, what you should do about it.

167
00:08:46.270 --> 00:08:50.200
<v James Loach>OK, um, and so you've got, you've got the
box in the middle that's taking that input

168
00:08:50.200 --> 00:08:52.419
<v James Loach>and producing that, that, that output.

169
00:08:53.000 --> 00:08:56.619
<v James Loach>Um, I guess we were going to this a little
bit, but Sensii is, um, a whole complex

170
00:08:56.619 --> 00:08:58.619
<v James Loach>arrangement of different things in that
box.

171
00:08:58.960 --> 00:09:00.299
<v James Loach>But you know, in.

172
00:09:00.840 --> 00:09:04.409
<v James Loach>Uh, simple terms, again, in, in an
old-style system, you could imagine that box

173
00:09:04.409 --> 00:09:05.820
<v James Loach>being full of rules, right?

174
00:09:06.090 --> 00:09:09.250
<v James Loach>And I think a lot of maintenance stuff
was, was really done, done like this, you

175
00:09:09.250 --> 00:09:12.929
<v James Loach>know, if the, the, the value of the
current crosses a a certain value, then it

176
00:09:12.929 --> 00:09:15.109
<v James Loach>probably means this or this sort of thing,
right?

177
00:09:15.510 --> 00:09:19.210
<v James Loach>Um, and then on the other extreme, you can
imagine having that as a pure machine

178
00:09:19.210 --> 00:09:20.130
<v James Loach>learning system.

179
00:09:20.369 --> 00:09:24.570
<v James Loach>Again, so maybe you, in the limits, you
don't know what's going on inside it, but it

180
00:09:24.570 --> 00:09:28.890
<v James Loach>just reliably and usefully maps that
contextual information onto statements or

181
00:09:28.890 --> 00:09:29.809
<v James Loach>actions, uh.

182
00:09:30.809 --> 00:09:31.969
<v James Loach>Advice, you know.

183
00:09:33.909 --> 00:09:34.530
<v Liz McGinn>Nice.

184
00:09:36.119 --> 00:09:37.150
<v Liz McGinn>That's interesting to hear.

185
00:09:37.200 --> 00:09:38.059
<v Liz McGinn>I mean, it's um.

186
00:09:38.969 --> 00:09:43.349
<v Liz McGinn>Definitely a lot of people go about the
rules in predictive maintenance, not quite

187
00:09:43.380 --> 00:09:48.690
<v Liz McGinn>predictive, I guess, but prescript, not
prescriptive, sorry, in the way that they

188
00:09:48.690 --> 00:09:51.590
<v Liz McGinn>are looking at it and they don't want it
to go over these thresholds.

189
00:09:51.849 --> 00:09:56.450
<v Liz McGinn>Um, so when we're using, say, machine
learning or AI, how does that train, like, I

190
00:09:56.450 --> 00:10:01.489
<v Liz McGinn>know you have this box that's giving you
different rules, but why, um, is that

191
00:10:01.489 --> 00:10:02.140
<v Liz McGinn>approach?

192
00:10:03.700 --> 00:10:04.820
<v Liz McGinn>Different or better, I guess.

193
00:10:04.859 --> 00:10:08.200
<v Liz McGinn>I don't want to use the term better, but
I'm kind of thinking better in my head.

194
00:10:08.429 --> 00:10:08.690
<v Liz McGinn>Um,

195
00:10:08.739 --> 00:10:13.090
<v James Loach>well, it, it, it, it's just very
complicated, um, answer that question because you,

196
00:10:13.140 --> 00:10:18.280
<v James Loach>you cannot at, at present, you know,
certainly for, um, the sort of very, um,

197
00:10:18.289 --> 00:10:22.940
<v James Loach>machine agnostic application like Sensi,
you cannot train a machine learning model

198
00:10:22.940 --> 00:10:28.700
<v James Loach>that can just convert that arbitrary set
of input information into, into, into

199
00:10:28.700 --> 00:10:32.280
<v James Loach>reliable output, you know, and do so
quickly and, and at scale.

200
00:10:33.119 --> 00:10:33.130
<v James Loach>Right?

201
00:10:34.109 --> 00:10:38.229
<v James Loach>So what, what you do inside that system,
um.

202
00:10:39.640 --> 00:10:44.789
<v James Loach>Yeah, I'm not sure how deeply to go into
sense I hear, right, but you, um, yeah, so

203
00:10:44.789 --> 00:10:48.090
<v James Loach>we have, we have a complex arrangement of
things, um.

204
00:10:49.049 --> 00:10:50.409
<v James Loach>Maybe we should start this one again.

205
00:10:50.690 --> 00:10:54.299
<v James Loach>What, what's, I suppose, what, what
specifically do you want me to?

206
00:10:54.650 --> 00:10:55.000
<v Liz McGinn>You're good.

207
00:10:55.049 --> 00:11:00.369
<v Liz McGinn>I was just curious, I'm kind of leaning
into the differences and kind of the benefit

208
00:11:00.369 --> 00:11:05.250
<v Liz McGinn>of using AI or machine learning, uh, in
comparison to kind of that traditional

209
00:11:05.250 --> 00:11:06.369
<v Liz McGinn>rule-based following.

210
00:11:06.440 --> 00:11:06.809
<v Liz McGinn>Yeah, OK,

211
00:11:06.929 --> 00:11:07.530
<v James Loach>OK, OK, OK.

212
00:11:07.729 --> 00:11:07.950
<v James Loach>OK.

213
00:11:10.369 --> 00:11:14.429
<v Liz McGinn>So when using AI and machine learning to
for predictive maintenance, how does that,

214
00:11:14.450 --> 00:11:19.679
<v Liz McGinn>um, Change the the viewpoint of kind of
this rule-based following, so the form of,

215
00:11:19.690 --> 00:11:24.450
<v Liz McGinn>um, you know, kind of that black box being
filled of questions and answers versus

216
00:11:24.450 --> 00:11:25.010
<v Liz McGinn>machine learning.

217
00:11:26.369 --> 00:11:30.650
<v James Loach>OK, so, so I think you think this is to do
with optimal use of information.

218
00:11:31.049 --> 00:11:31.530
<v James Loach>OK.

219
00:11:31.809 --> 00:11:36.510
<v James Loach>Um, so, you know, if, if you've got some
rules, I mean, somebody has to have, have

220
00:11:36.770 --> 00:11:40.489
<v James Loach>come up with these and they, they might be
sort of sensible rules, but they, they

221
00:11:40.489 --> 00:11:44.770
<v James Loach>are, they are quite crude things, you
know, like if it's above a certain value, it

222
00:11:44.770 --> 00:11:45.130
<v James Loach>means this.

223
00:11:45.169 --> 00:11:48.530
<v James Loach>If it's below, um, it, it, it's not a
problem, right?

224
00:11:48.739 --> 00:11:53.289
<v James Loach>But you know you can imagine the data has
all kinds of values, and you may see the

225
00:11:53.289 --> 00:11:57.619
<v James Loach>initial indications of some problem
developing before it crosses the line.

226
00:11:57.979 --> 00:12:01.159
<v James Loach>And then even if it's crossed that line,
you may see in the history of the asset

227
00:12:01.159 --> 00:12:03.840
<v James Loach>that actually if it's only just a little
bit above it, it doesn't matter.

228
00:12:04.679 --> 00:12:08.359
<v James Loach>But alternatively, in another situation,
it was just a little bit above, that really

229
00:12:08.359 --> 00:12:10.159
<v James Loach>is a really big problem and what have you.

230
00:12:10.489 --> 00:12:13.570
<v James Loach>So, I mean, it's just a question, you have
a lot of complexity and subtlety in the

231
00:12:13.570 --> 00:12:18.489
<v James Loach>real world and you are representing that
in a rules-based system as simple rules

232
00:12:18.489 --> 00:12:22.250
<v James Loach>that a human can understand, but there,
there are abstractions and simplifications

233
00:12:22.250 --> 00:12:23.070
<v James Loach>of the world.

234
00:12:23.409 --> 00:12:27.340
<v James Loach>And I guess the, the promise of machine
learning would, would be that you can by

235
00:12:27.340 --> 00:12:30.729
<v James Loach>learning, um, based on all the, all the
things that have happened and all that you

236
00:12:30.729 --> 00:12:31.130
<v James Loach>know.

237
00:12:31.460 --> 00:12:34.219
<v James Loach>And that you can just do this in a much
more subtle way, right?

238
00:12:34.429 --> 00:12:39.390
<v James Loach>Picking up on all the little, um, little
changes in the data, all the historical

239
00:12:39.390 --> 00:12:43.469
<v James Loach>things and just very optimally say, OK, I
think now is the day you need to go and

240
00:12:43.469 --> 00:12:45.750
<v James Loach>have a look at that for, for these
reasons.

241
00:12:46.239 --> 00:12:49.880
<v James Loach>Yeah, so you, you do, you do that more in
a, in a more optimized way.

242
00:12:49.960 --> 00:12:53.239
<v James Loach>You make better use of the information,
and this means you make better use of the

243
00:12:53.239 --> 00:12:55.659
<v James Loach>user's time and, and waste their timeless.

244
00:12:55.960 --> 00:12:57.469
<v James Loach>Yeah, that would be the promise, right?

245
00:12:57.719 --> 00:13:02.419
<v James Loach>But in, in reality, as I said, you, it's
difficult at the moment to build this pure,

246
00:13:02.619 --> 00:13:08.539
<v James Loach>um, Machine learning system to do that,
but everything is moving in that direction

247
00:13:08.679 --> 00:13:14.960
<v James Loach>in a sense like our system is a complex
mix of AIML things and statistical things,

248
00:13:15.159 --> 00:13:15.580
<v Liz McGinn>right?

249
00:13:16.280 --> 00:13:16.289
<v Liz McGinn>Mhm.

250
00:13:17.429 --> 00:13:22.390
<v Liz McGinn>Yeah, plus human input is important for, I
guess, the machine learning as well for

251
00:13:22.390 --> 00:13:23.400
<v Liz McGinn>our systems.

252
00:13:24.640 --> 00:13:28.320
<v James Loach>Yeah, and that's super important and again
something something we go on to, but you

253
00:13:28.320 --> 00:13:31.919
<v James Loach>know the system really, really wants,
wants, wants to know about the, about the

254
00:13:31.919 --> 00:13:36.000
<v James Loach>world, right, how it was in the past and
how it, how it is now and doesn't have

255
00:13:36.000 --> 00:13:40.239
<v James Loach>direct access to that, you know, you have
access through things that users write or

256
00:13:40.239 --> 00:13:44.559
<v James Loach>time series that you collect or maybe a
photo that the user takes or something like

257
00:13:44.559 --> 00:13:44.789
<v James Loach>that.

258
00:13:45.039 --> 00:13:49.159
<v James Loach>But at the moment, and this may be
different in future, the user is required, right?

259
00:13:49.429 --> 00:13:49.919
<v James Loach>It's.

260
00:13:50.770 --> 00:13:55.450
<v James Loach>The entity that really tells you what
these things in the data mean, right, so you

261
00:13:55.450 --> 00:13:59.650
<v James Loach>can learn what they mean and, you know, to
do this task in the future, yeah,

262
00:14:00.460 --> 00:14:01.859
<v Liz McGinn>yeah, because it's always learning.

263
00:14:03.369 --> 00:14:06.340
<v James Loach>Yeah, yeah, yeah, but it has to get, it
has to get ground truth knowledge from

264
00:14:06.340 --> 00:14:08.820
<v James Loach>somewhere and Senii is, is.

265
00:14:09.580 --> 00:14:12.059
<v James Loach>Works around the user because the user
normally knows, right?

266
00:14:12.150 --> 00:14:15.419
<v James Loach>They're in the factory, they've been
working on this machine for 3 years or what

267
00:14:15.419 --> 00:14:15.929
<v James Loach>have you.

268
00:14:16.210 --> 00:14:17.380
<v James Loach>They have some tacit knowledge.

269
00:14:17.559 --> 00:14:21.580
<v James Loach>You, you want to get that information out
their head and get it into the system and

270
00:14:21.580 --> 00:14:24.760
<v James Loach>then use that knowledge, you know, while
the user is doing something else.

271
00:14:26.419 --> 00:14:30.190
<v Liz McGinn>Yeah, so do you think um AI will ever kind
of replace that human or do you think a

272
00:14:30.190 --> 00:14:33.539
<v Liz McGinn>human will always be needed for this kind
of thing?

273
00:14:36.349 --> 00:14:41.900
<v James Loach>Um, it, well, if, if you, you say always,
I think, I think the answer is, is no.

274
00:14:42.270 --> 00:14:43.700
<v James Loach>Um, I mean, it, it.

275
00:14:44.340 --> 00:14:50.400
<v James Loach>Because it depends on your worldview at,
at, at some level, but I, I would see human

276
00:14:50.400 --> 00:14:53.719
<v James Loach>beings as fundamentally machine-like, OK?

277
00:14:54.130 --> 00:14:58.859
<v James Loach>And I don't see any particular reason why
anything that human beings do can't be,

278
00:14:58.979 --> 00:15:04.239
<v James Loach>can't be replicated artificially, um, and
then ultimately done a lot better.

279
00:15:04.859 --> 00:15:05.419
<v James Loach>OK?

280
00:15:05.669 --> 00:15:10.359
<v James Loach>Um, and then, you know, if, if you're in
that situation where you can very cheaply.

281
00:15:11.080 --> 00:15:15.169
<v James Loach>Um, and accurately to do all of these kind
of things with a robot.

282
00:15:15.479 --> 00:15:18.969
<v James Loach>Um, at that point, humans would be doing
it as a hobby, I would think.

283
00:15:19.380 --> 00:15:20.580
<v James Loach>OK, right?

284
00:15:20.969 --> 00:15:25.039
<v James Loach>Um, much like we, we continue to play
chess even though, uh, you know, chess

285
00:15:25.039 --> 00:15:27.210
<v James Loach>computers can, can be even the best chess
players.

286
00:15:27.559 --> 00:15:31.429
<v James Loach>But anyway, but, but that, that could be a
very, very long way, way away.

287
00:15:31.719 --> 00:15:32.059
<v James Loach>Yeah.

288
00:15:32.400 --> 00:15:37.400
<v James Loach>And where you have, you know, expensive,
imperfect AI systems, obviously good

289
00:15:37.400 --> 00:15:40.419
<v James Loach>reasons to keep, keep human beings
involved.

290
00:15:41.080 --> 00:15:47.030
<v James Loach>And even though you can kind of see a
fuzzy route to complete a replacement of Um,

291
00:15:47.150 --> 00:15:50.070
<v James Loach>manual labor and knowledge work like our
jobs also, right?

292
00:15:50.080 --> 00:15:55.150
<v James Loach>You can kind of see a fuzzy route to that,
you know, I think breakthroughs would be

293
00:15:55.150 --> 00:15:59.030
<v James Loach>needed and the timescale for this to
happen is still quite, quite uncertain and

294
00:15:59.030 --> 00:15:59.890
<v James Loach>could be long, yeah.

295
00:16:01.159 --> 00:16:02.799
<v Liz McGinn>Well, I'm glad it's got a minute.

296
00:16:02.840 --> 00:16:05.820
<v Liz McGinn>I have to reach retirement before they
start replacing my job, hopefully.

297
00:16:09.340 --> 00:16:16.109
<v Liz McGinn>Um So do you think that um when you go
back to the, the manufacturing, like,

298
00:16:16.159 --> 00:16:20.080
<v Liz McGinn>vision of it, I know that it's kind of
hard to talk to um people like are resistant

299
00:16:20.080 --> 00:16:24.979
<v Liz McGinn>to change, similar to like kind of
manufacturers are kind of usually slower to, uh,

300
00:16:25.210 --> 00:16:28.900
<v Liz McGinn>you know, move towards the direction, but
I'm seeing a lot of changes in that, um,

301
00:16:29.010 --> 00:16:34.570
<v Liz McGinn>not like, quickly, but like they are kind
of leaning into this AI transformation.

302
00:16:34.650 --> 00:16:41.619
<v Liz McGinn>Why do you think um this one's being Like
Um, except it's easier,

303
00:16:42.320 --> 00:16:42.919
<v Liz McGinn>or is it not?

304
00:16:43.039 --> 00:16:43.820
<v Liz McGinn>Am I just,

305
00:16:44.280 --> 00:16:46.750
<v James Loach>well, I, I think people are smart, right?

306
00:16:46.880 --> 00:16:52.840
<v James Loach>You know, and when, I mean, the, the
current AI systems based on language models are

307
00:16:52.840 --> 00:16:54.010
<v James Loach>really very useful.

308
00:16:54.169 --> 00:16:58.099
<v James Loach>And, uh, yeah, I mean, I, I use them all
day every day.

309
00:16:58.479 --> 00:17:03.000
<v James Loach>Uh, I expect most people listening to this
probably to also, or at least a, a good

310
00:17:03.000 --> 00:17:04.140
<v James Loach>chunk of time.

311
00:17:04.599 --> 00:17:08.260
<v James Loach>Um, when you have something that is real
like that and so obviously useful.

312
00:17:08.670 --> 00:17:12.910
<v James Loach>Um, yeah, it's, it's no surprise people
are interested adopt in adopting it and

313
00:17:12.910 --> 00:17:17.699
<v James Loach>working how you take that sort of raw
power and intelligence and somehow get it to

314
00:17:17.699 --> 00:17:21.150
<v James Loach>work, you know, properly in your
particular environment, if that's your, your

315
00:17:21.150 --> 00:17:22.689
<v James Loach>factory or something like that.

316
00:17:23.150 --> 00:17:23.729
<v James Loach>OK?

317
00:17:24.150 --> 00:17:24.589
<v James Loach>Um.

318
00:17:25.630 --> 00:17:31.199
<v James Loach>You might contrast that to before language
models where machine learning was still

319
00:17:31.199 --> 00:17:36.760
<v James Loach>used in factories for all kinds of tasks,
but this is more narrow AI, so doing like

320
00:17:36.760 --> 00:17:41.119
<v James Loach>very specific things and you have to go
through, you know, maybe long expensive

321
00:17:41.119 --> 00:17:45.760
<v James Loach>training processes and keep rebuilding,
you know, and have data scientists involved

322
00:17:45.760 --> 00:17:47.180
<v James Loach>all the time, this kind of thing.

323
00:17:48.050 --> 00:17:51.969
<v James Loach>Yeah, so even though there was something
there that, that, that was working and you

324
00:17:51.969 --> 00:17:58.329
<v James Loach>could show off demos, yeah, very expensive
in general, and the reality of those

325
00:17:58.329 --> 00:18:02.829
<v James Loach>systems is often something of a
disappointment, whereas most people's interaction

326
00:18:02.829 --> 00:18:05.329
<v James Loach>with language model-based eye is kind of.

327
00:18:05.609 --> 00:18:09.790
<v James Loach>It's wearing off a bit now, but some kind
of awe, it's like magic, yeah, and so

328
00:18:10.010 --> 00:18:11.819
<v James Loach>it's, it's a very different headspace,

329
00:18:11.829 --> 00:18:12.869
<v Liz McGinn>yeah, I think.

330
00:18:13.089 --> 00:18:13.319
<v Liz McGinn>Yeah.

331
00:18:13.489 --> 00:18:18.449
<v Liz McGinn>So what did it look like before we went
into kind of this AI, like the more basic

332
00:18:18.449 --> 00:18:23.489
<v Liz McGinn>version of what you would see like in
manufacturing for, um, those kind of questions

333
00:18:23.489 --> 00:18:23.969
<v Liz McGinn>and answers?

334
00:18:24.010 --> 00:18:25.489
<v Liz McGinn>What were they using it for in there?

335
00:18:28.219 --> 00:18:32.349
<v James Loach>Yeah, I mean, I suppose you could imagine
things like spotting defects, something

336
00:18:32.349 --> 00:18:32.790
<v James Loach>like that.

337
00:18:32.949 --> 00:18:37.380
<v James Loach>So you maybe have some vision system and
you have lots of examples.

338
00:18:37.589 --> 00:18:42.550
<v James Loach>This is a good thing and this is a bad
thing, you know, different categories of bad

339
00:18:42.550 --> 00:18:47.109
<v James Loach>thing, and you know, you collect all of
these images and you have to then, you have

340
00:18:47.109 --> 00:18:50.910
<v James Loach>to train a model and you've got to
commission it and then you have the problems if

341
00:18:50.910 --> 00:18:54.069
<v James Loach>the lighting slightly changes in your
factory, and you've got to be resistant to

342
00:18:54.069 --> 00:18:56.709
<v James Loach>that and you're doing all that kind of
thing, yeah.

343
00:18:57.030 --> 00:19:02.390
<v James Loach>And then, yeah, the, the reason it's more
exciting with systems based on language

344
00:19:02.390 --> 00:19:08.910
<v James Loach>models is, yeah, like you log into such a
model and take a photo of your product and

345
00:19:08.910 --> 00:19:13.550
<v James Loach>ask it whether it looks OK, and the system
immediately has some opinion on that,

346
00:19:13.589 --> 00:19:14.849
<v James Loach>which is quite robust.

347
00:19:15.400 --> 00:19:19.479
<v James Loach>You know, it doesn't replace that previous
system, but, but there's the sort of

348
00:19:19.479 --> 00:19:21.640
<v James Loach>power there that you get, get for free,
right?

349
00:19:21.800 --> 00:19:26.239
<v James Loach>It feels like it's for free, and then it's
like trying to, to tame this beast, you

350
00:19:26.239 --> 00:19:29.520
<v James Loach>know, whereas to me, like traditional
machine learning systems is like, you know,

351
00:19:29.560 --> 00:19:35.319
<v James Loach>climbing big mountains is very hard, you
know, um, yeah, to, to, yeah, to make these

352
00:19:35.319 --> 00:19:38.260
<v James Loach>things work and be robust and what have
you, like a hard grind.

353
00:19:38.359 --> 00:19:41.069
<v James Loach>You don't feel you're getting free magic
from the world somehow.

354
00:19:42.209 --> 00:19:46.089
<v Liz McGinn>Yeah, you have to keep correcting it
pretty much anytime something's a little bit

355
00:19:46.089 --> 00:19:46.609
<v Liz McGinn>different.

356
00:19:48.099 --> 00:19:50.400
<v James Loach>Or making it robust, yeah, all that kind
of stuff.

357
00:19:52.130 --> 00:19:52.849
<v Liz McGinn>Nice.

358
00:19:53.550 --> 00:19:59.380
<v Liz McGinn>Um, So, when we're talking about machine
learning and AI we're adding a bunch of

359
00:19:59.380 --> 00:20:06.239
<v Liz McGinn>data into it and I know our data is
trained on historical data um and live data, but

360
00:20:06.780 --> 00:20:11.780
<v Liz McGinn>what happens if you don't have enough of
that historical data to really build a

361
00:20:11.780 --> 00:20:12.219
<v Liz McGinn>model?

362
00:20:13.040 --> 00:20:13.050
<v Liz McGinn>OK

363
00:20:15.199 --> 00:20:17.239
<v James Loach>Yeah, again, this, this is an interesting
thing.

364
00:20:17.400 --> 00:20:23.250
<v James Loach>So like old style or I guess you'd say
more, more traditional systems, if you, if

365
00:20:23.250 --> 00:20:28.479
<v James Loach>you want to build a machine learning model
to work with your, your assets and you've

366
00:20:28.479 --> 00:20:30.319
<v James Loach>got no data, right?

367
00:20:30.560 --> 00:20:34.319
<v James Loach>You know, like, it's a bit difficult to
understand what you, what, what, what, what

368
00:20:34.319 --> 00:20:36.680
<v James Loach>you imagine, what one would imagine
happening at that point, right?

369
00:20:36.880 --> 00:20:40.959
<v James Loach>Like you just have to take a model from
somewhere else or try and get data from

370
00:20:40.959 --> 00:20:42.800
<v James Loach>somewhere or you have to collect the data.

371
00:20:43.000 --> 00:20:43.859
<v James Loach>It can't be sort of.

372
00:20:44.650 --> 00:20:46.790
<v James Loach>Magicked out of nothing.

373
00:20:47.750 --> 00:20:53.459
<v James Loach>Um, again, like language models, you know,
they've read, read the entire internet,

374
00:20:53.489 --> 00:20:58.839
<v James Loach>um, and synthesized all that information,
and they come with reasonable, with a

375
00:20:58.839 --> 00:21:02.060
<v James Loach>reasonable sort of prior understanding of
many situations.

376
00:21:02.599 --> 00:21:07.660
<v James Loach>Um, so, you know, a language model system,
if you, you know, tell it you have a

377
00:21:08.000 --> 00:21:13.040
<v James Loach>robot of a particular type, you know, um,
it's going to have some built inbuilt

378
00:21:13.040 --> 00:21:16.300
<v James Loach>understanding of robots and what happens
to robots, um.

379
00:21:16.869 --> 00:21:19.790
<v James Loach>And maybe your particular model, right?

380
00:21:19.910 --> 00:21:23.650
<v James Loach>And you can very easily supplement that
with a, a few facts and some extra

381
00:21:23.650 --> 00:21:25.300
<v James Loach>information, um, and what have you.

382
00:21:25.670 --> 00:21:29.670
<v James Loach>And, and, and you can again, very, very
cheaply and easily have something which

383
00:21:29.670 --> 00:21:33.410
<v James Loach>could immediately start to do something
useful, right, when you don't have any data

384
00:21:33.829 --> 00:21:39.589
<v James Loach>to build these more traditional, um, you
know, predictive models or digital twins or

385
00:21:39.589 --> 00:21:40.349
<v James Loach>things like that.

386
00:21:40.969 --> 00:21:44.569
<v James Loach>So again, you, you know, you have this,
you, you're not starting from zero and

387
00:21:44.569 --> 00:21:46.689
<v James Loach>building up and trying to get the
information to do that.

388
00:21:46.810 --> 00:21:50.530
<v James Loach>You, you have this kind of fuzzy, slightly
problematic in certain ways, but, you

389
00:21:50.530 --> 00:21:55.010
<v James Loach>know, it's, um, great big world model and
you, you're taking that and then, you

390
00:21:55.010 --> 00:21:58.770
<v James Loach>know, um, in certain of these systems,
you'll be taking that and you'd be

391
00:21:58.770 --> 00:22:01.829
<v James Loach>customizing it to your particular problem,
directing it in the right way.

392
00:22:02.589 --> 00:22:07.130
<v James Loach>Yeah, so, so it, it, it does allow you,
and this is hopefully something we'll see in

393
00:22:07.130 --> 00:22:13.069
<v James Loach>Sensai at some point, get nice prognostic
output that's sensible and useful, really

394
00:22:13.069 --> 00:22:16.229
<v James Loach>with sort of very little input
information, right?

395
00:22:16.469 --> 00:22:21.430
<v James Loach>And then just have those, um, predictions
just be sort of refined as you start to

396
00:22:21.430 --> 00:22:25.709
<v James Loach>collect data and, you know, more knowledge
about your machine, yeah.

397
00:22:27.000 --> 00:22:27.319
<v Liz McGinn>Yeah.

398
00:22:27.540 --> 00:22:31.599
<v Liz McGinn>So when we're talking about Sensei, is it
starting all on kind of that baseline

399
00:22:31.599 --> 00:22:37.479
<v Liz McGinn>model that's already been built um by your
team, obviously, and then adapting it per

400
00:22:37.479 --> 00:22:39.280
<v Liz McGinn>customer when they get their data in
there?

401
00:22:39.329 --> 00:22:40.780
<v Liz McGinn>Is it all kind of the same?

402
00:22:42.709 --> 00:22:43.760
<v Liz McGinn>Box that you're putting in.

403
00:22:44.510 --> 00:22:48.709
<v James Loach>So since I worked in a very
non-traditional way, I, I think, like even, even before

404
00:22:48.709 --> 00:22:53.890
<v James Loach>generative AI, um, you know, you know, we
don't have persistent models of machines.

405
00:22:54.270 --> 00:22:54.709
<v James Loach>OK?

406
00:22:54.989 --> 00:22:57.500
<v James Loach>You know, we were taking in streams of
data from machines.

407
00:22:57.790 --> 00:23:01.630
<v James Loach>Yeah, we have a whole series of
traditional machine learning and statistical

408
00:23:01.630 --> 00:23:06.069
<v James Loach>pipelines, which are looking at that, that
data for anything that, that, that might

409
00:23:06.069 --> 00:23:06.729
<v James Loach>be of interest.

410
00:23:07.000 --> 00:23:11.550
<v James Loach>So some unusual trends or some spikes or
odd forecast or a match or something like

411
00:23:11.550 --> 00:23:11.869
<v James Loach>this.

412
00:23:12.449 --> 00:23:12.709
<v James Loach>OK.

413
00:23:12.859 --> 00:23:15.760
<v James Loach>So you're just, just making these
observations in quite a standard way.

414
00:23:16.060 --> 00:23:19.780
<v James Loach>And those systems when they're using
models, those models are typically built on the

415
00:23:19.780 --> 00:23:20.680
<v James Loach>fly, right?

416
00:23:20.770 --> 00:23:24.729
<v James Loach>So you get the data and builds a model of
the data and then applies that model to

417
00:23:24.729 --> 00:23:26.780
<v James Loach>recent data to look for unusual things.

418
00:23:27.260 --> 00:23:27.609
<v James Loach>OK?

419
00:23:27.819 --> 00:23:32.280
<v James Loach>So it's like, since in that part of our
system is like a universal modeling system.

420
00:23:32.689 --> 00:23:37.060
<v James Loach>It's the same system that's modeling all
data and you get a, a custom.

421
00:23:37.550 --> 00:23:42.109
<v James Loach>ephemeral model that you make and throw
away every single time point custom for

422
00:23:42.109 --> 00:23:43.089
<v James Loach>every single asset.

423
00:23:44.449 --> 00:23:48.859
<v James Loach>OK, so we know and we don't have like a
model of a robot stored anywhere.

424
00:23:49.819 --> 00:23:52.140
<v James Loach>And we model the data as it, as it comes
in, yeah.

425
00:23:52.780 --> 00:23:57.089
<v James Loach>And then the way, the way we get this to
work, um, well, this is very, a very nice

426
00:23:57.089 --> 00:24:00.260
<v James Loach>thing to have because, you know, Sensor is
supposed to be a, a general-purpose

427
00:24:00.260 --> 00:24:02.780
<v James Loach>system go with any data from any assets,
right?

428
00:24:03.020 --> 00:24:07.900
<v James Loach>And, and, you know, um, one way to do this
is really to treat the data just as it

429
00:24:07.900 --> 00:24:10.260
<v James Loach>is, treat everything as individual in
particular.

430
00:24:10.579 --> 00:24:10.890
<v James Loach>OK?

431
00:24:11.060 --> 00:24:13.689
<v James Loach>So we take this data and we're putting all
these ephemeral models and we're just

432
00:24:13.689 --> 00:24:17.300
<v James Loach>looking for things of potential interest,
and then we store those and we note those

433
00:24:17.300 --> 00:24:17.640
<v James Loach>down.

434
00:24:18.439 --> 00:24:21.680
<v James Loach>Yeah, then we have like a trigger system
which analyzes those.

435
00:24:21.959 --> 00:24:25.880
<v James Loach>So it's looking, oh, what things have been
spotted on this asset recently, and then

436
00:24:25.880 --> 00:24:30.439
<v James Loach>they look at similar assets, similar
situations, and then it tries to predict

437
00:24:30.439 --> 00:24:32.800
<v James Loach>whether you're going to care as the user.

438
00:24:33.729 --> 00:24:38.239
<v James Loach>OK, and then if it thinks you're likely to
care, this gets passed to our generative

439
00:24:38.239 --> 00:24:42.829
<v James Loach>AI system which analyzes that trigger,
what's going on, what's happened in the past,

440
00:24:43.689 --> 00:24:47.209
<v James Loach>looks up in the documents, looks at the
deep history of the asset, all this kind of

441
00:24:47.209 --> 00:24:51.890
<v James Loach>thing, and says, right, this is what's
going on, this is what it could, could be.

442
00:24:53.420 --> 00:24:57.750
<v James Loach>OK, so that whole architecture is, is
really quite general and nowhere in there are

443
00:24:57.750 --> 00:24:59.790
<v James Loach>we treating any asset specially.

444
00:25:00.270 --> 00:25:02.810
<v James Loach>An asset to us is just a collection of
time series.

445
00:25:03.510 --> 00:25:07.420
<v James Loach>You know, that has work events and
feedback and notes and stuff like that,

446
00:25:07.719 --> 00:25:07.949
<v Liz McGinn>yeah.

447
00:25:08.859 --> 00:25:13.979
<v Liz McGinn>Yeah, OK, so it's just collecting um and
creating models constantly until it

448
00:25:13.979 --> 00:25:18.699
<v Liz McGinn>triggers something that'll kick it over to
the Gen AI portion of it, where then

449
00:25:18.699 --> 00:25:19.280
<v Liz McGinn>it'll.

450
00:25:20.290 --> 00:25:23.790
<v Liz McGinn>Try to figure out based on other context
of what's going on.

451
00:25:24.589 --> 00:25:24.930
<v James Loach>Yeah.

452
00:25:25.270 --> 00:25:26.910
<v James Loach>And that architecture works very well,
right?

453
00:25:26.989 --> 00:25:32.270
<v James Loach>And it's very important for sense size,
um, ability to scale, right, and be asset

454
00:25:32.270 --> 00:25:36.819
<v James Loach>agnostic, but it, it is a bit unusual,
right, for asset, for, for, for applications

455
00:25:36.819 --> 00:25:37.510
<v James Loach>of this type.

456
00:25:37.989 --> 00:25:40.910
<v James Loach>Yeah, and many people, yeah, will talk
about things in, in the terms that you asked

457
00:25:40.910 --> 00:25:41.650
<v James Loach>the question, right?

458
00:25:41.910 --> 00:25:45.229
<v James Loach>You know, I've got my robot, let's build a
model of the robot and store it in the

459
00:25:45.229 --> 00:25:48.229
<v James Loach>model's database, and now we've got the
data, we'll get the model and we'll do that

460
00:25:48.229 --> 00:25:48.750
<v James Loach>stuff, right?

461
00:25:48.910 --> 00:25:49.739
<v James Loach>We do not do that.

462
00:25:49.800 --> 00:25:49.810
<v Liz McGinn>Yeah.

463
00:25:50.589 --> 00:25:51.520
<v Liz McGinn>Interesting.

464
00:25:52.030 --> 00:25:53.170
<v Liz McGinn>So it's just like a bunch of.

465
00:25:53.989 --> 00:25:55.130
<v Liz McGinn>Repeated ephemeral.

466
00:25:57.420 --> 00:25:59.079
<v Liz McGinn>So the ephemeral is the throwaway.

467
00:25:59.229 --> 00:26:01.099
<v Liz McGinn>That's what you're referencing when you're
saying that.

468
00:26:01.219 --> 00:26:01.560
<v Liz McGinn>OK.

469
00:26:01.819 --> 00:26:02.780
<v Liz McGinn>Oh, interesting.

470
00:26:03.849 --> 00:26:06.349
<v Liz McGinn>That feels like it feels very complicated.

471
00:26:07.699 --> 00:26:12.489
<v James Loach>No, it's got the, the diagram has lots of
boxes and arrows on it, but it, it's

472
00:26:12.489 --> 00:26:14.219
<v James Loach>fundamentally not, not complicated.

473
00:26:14.339 --> 00:26:15.660
<v James Loach>It's those three bits,

474
00:26:15.819 --> 00:26:16.160
<v Liz McGinn>yeah,

475
00:26:16.660 --> 00:26:17.540
<v James Loach>um, and then buried

476
00:26:17.540 --> 00:26:19.400
<v Liz McGinn>in that, yeah, yeah.

477
00:26:19.780 --> 00:26:21.739
<v Liz McGinn>You say that as the expert, but I'm like,
wow.

478
00:26:22.579 --> 00:26:26.140
<v James Loach>So, and, and then in that there are all
kinds of machine learning systems for doing

479
00:26:26.140 --> 00:26:26.939
<v James Loach>different things, right?

480
00:26:27.099 --> 00:26:32.739
<v James Loach>So we have a Gen AI machine learning, uh,
system for forecasting time series, which

481
00:26:32.739 --> 00:26:35.229
<v James Loach>is super cool and absolute best in class.

482
00:26:35.599 --> 00:26:37.839
<v James Loach>Um, you know, and then you have lots of
minor little systems.

483
00:26:37.920 --> 00:26:41.719
<v James Loach>So when you spot something in a time
series, we have like a classifier system which

484
00:26:41.719 --> 00:26:44.829
<v James Loach>will look at that and say, oh, that's a
little upward trend or that's a downward

485
00:26:44.829 --> 00:26:47.000
<v James Loach>spike or something like that, you know.

486
00:26:47.520 --> 00:26:49.550
<v James Loach>So it's just, yeah, anyway, OK.

487
00:26:49.910 --> 00:26:52.780
<v Liz McGinn>A bunch of machine learning is all wrapped
in one to AI.

488
00:26:54.079 --> 00:26:54.459
<v Liz McGinn>Yeah.

489
00:26:55.000 --> 00:26:55.060
<v Liz McGinn>Nice.

490
00:26:56.890 --> 00:26:57.660
<v Liz McGinn>That's interesting.

491
00:26:57.790 --> 00:26:59.520
<v Liz McGinn>So, we've talked a little bit how it is
now.

492
00:26:59.859 --> 00:27:02.760
<v Liz McGinn>I guess, what future do you see?

493
00:27:02.839 --> 00:27:07.310
<v Liz McGinn>you said a couple of fuzzy things, um,
that'll take, you know, some time to get in

494
00:27:07.310 --> 00:27:11.109
<v Liz McGinn>there, but what do you see kind of on the
horizon that's Newer

495
00:27:14.489 --> 00:27:21.369
<v James Loach>Um Yeah, so I think in the, in the near
term, you know, what

496
00:27:21.369 --> 00:27:25.050
<v James Loach>we're interested in doing is getting a
hold of more information, more context,

497
00:27:25.170 --> 00:27:30.050
<v James Loach>real-world stuff about what is actually
happening, um, and that can come from taking

498
00:27:30.050 --> 00:27:34.410
<v James Loach>in new sources of data or connecting to,
you know, for example, other customer

499
00:27:34.410 --> 00:27:36.119
<v James Loach>systems or, or what have you.

500
00:27:36.489 --> 00:27:36.729
<v James Loach>Yeah.

501
00:27:37.050 --> 00:27:39.609
<v James Loach>And also accepting other kinds of input
from the user.

502
00:27:39.689 --> 00:27:43.290
<v James Loach>Maybe the user can upload photographs or
videos or things like that, but you just

503
00:27:43.290 --> 00:27:47.890
<v James Loach>want, you want more information about
what's happening in the world and you want a

504
00:27:47.890 --> 00:27:49.430
<v James Loach>smarter and more powerful.

505
00:27:50.420 --> 00:27:55.859
<v James Loach>You know, language model essentially, um,
to, to feed all of this stuff into, um,

506
00:27:56.020 --> 00:28:00.739
<v James Loach>and get, get more precise, um, and sort of
accurate output.

507
00:28:01.140 --> 00:28:01.339
<v James Loach>Yeah.

508
00:28:01.579 --> 00:28:04.109
<v James Loach>So, so we're interested in, in getting
more stuff in.

509
00:28:04.339 --> 00:28:07.260
<v James Loach>And then when it goes into the language
model, you know, we're interested in, in

510
00:28:07.260 --> 00:28:11.630
<v James Loach>using this stuff in most, you know, better
models and using more sophisticated ways,

511
00:28:11.739 --> 00:28:17.300
<v James Loach>so more agentic, more reasoning, more
thinking about stuff, um, and, and to get

512
00:28:17.300 --> 00:28:20.099
<v James Loach>higher quality answers, OK?

513
00:28:20.400 --> 00:28:24.150
<v James Loach>Um, and so, so, so in the near term,
there, there is plenty of mileage there and

514
00:28:24.150 --> 00:28:28.469
<v James Loach>there is, there is much to do and lots of
improvements in performance that you, that

515
00:28:28.469 --> 00:28:29.349
<v James Loach>you could imagine.

516
00:28:30.199 --> 00:28:33.880
<v James Loach>Um, yeah, so it's sort of longer term
after that, you, you can imagine, so, I said

517
00:28:33.880 --> 00:28:36.599
<v James Loach>that, you know, you've got your input,
you've got your box, and you've got, got your

518
00:28:36.599 --> 00:28:37.199
<v James Loach>output.

519
00:28:37.430 --> 00:28:39.520
<v James Loach>This box is full of stuff, right?

520
00:28:39.640 --> 00:28:44.959
<v James Loach>One can imagine simplifying this over,
over time, yeah, um, making it, um, you know,

521
00:28:45.119 --> 00:28:49.329
<v James Loach>removing statistical things, having more
machine learning, merging machine learning

522
00:28:49.329 --> 00:28:53.719
<v James Loach>systems, and trying to have it more of an
integrated intelligence rather than a

523
00:28:53.719 --> 00:28:55.939
<v James Loach>collection of lots of little things
working together.

524
00:28:56.689 --> 00:29:00.319
<v James Loach>That's an interesting thing from an
internal perspective.

525
00:29:01.280 --> 00:29:07.660
<v James Loach>Um, and then a little bit longer term
again, I suppose going back to Context, I

526
00:29:07.660 --> 00:29:12.819
<v James Loach>mean, you know, there will be a time,
maybe it's not too distant future where you

527
00:29:12.819 --> 00:29:15.760
<v James Loach>can have your, um, your humanoid robots,
right?

528
00:29:15.900 --> 00:29:20.040
<v James Loach>Working with maintenance engineers in the,
in the factories, OK?

529
00:29:20.420 --> 00:29:23.699
<v James Loach>Um, and that, this is, this is kind of
nice then, right?

530
00:29:23.939 --> 00:29:28.170
<v James Loach>Because, you know, rather than relying on
people to write things to you or take a

531
00:29:28.170 --> 00:29:29.560
<v James Loach>photo and send it to you.

532
00:29:29.969 --> 00:29:34.989
<v James Loach>If you have these characters on the scene
there where you can take the sensory

533
00:29:34.989 --> 00:29:38.569
<v James Loach>information in whilst they are looking at
the machine and eventually interacting in

534
00:29:38.569 --> 00:29:43.469
<v James Loach>the machine, you can have much tighter
feedback loops between the kind of background

535
00:29:43.469 --> 00:29:44.589
<v James Loach>intelligence.

536
00:29:45.579 --> 00:29:50.540
<v James Loach>And information, you know, integration
that Sensai has, yeah, and then on the scene,

537
00:29:50.979 --> 00:29:55.800
<v James Loach>you know, live kind of, um, knowledge
about what, what is actually happening, like

538
00:29:55.800 --> 00:29:57.000
<v James Loach>right, right, right on the machine.

539
00:29:57.380 --> 00:29:59.400
<v James Loach>So you sort of bring those things closer
together.

540
00:30:00.170 --> 00:30:03.410
<v James Loach>Um, yeah, and then there's, there's some
very long-term picture.

541
00:30:04.109 --> 00:30:08.829
<v James Loach>Um, yeah, where, where you, you have these
humanoid robots and Sensii is, um,

542
00:30:08.969 --> 00:30:12.599
<v James Loach>perhaps part of that story and, yeah,
directing them to what they should be doing

543
00:30:12.599 --> 00:30:16.420
<v James Loach>and learning from them and get some kind
of acceleration.

544
00:30:18.770 --> 00:30:22.390
<v Liz McGinn>Get a smarter, more context, better.

545
00:30:22.400 --> 00:30:22.930
<v Liz McGinn>Well, this is it, right?

546
00:30:23.010 --> 00:30:23.310
<v Liz McGinn>Yeah,

547
00:30:23.770 --> 00:30:26.130
<v James Loach>that's what rotry have is to get more
context, right?

548
00:30:26.369 --> 00:30:29.689
<v James Loach>When you, you get more context and you're
able to integrate that into your models

549
00:30:29.689 --> 00:30:33.130
<v James Loach>effectively, those models get smarter and
you're able to make better decisions next

550
00:30:33.130 --> 00:30:33.479
<v James Loach>time.

551
00:30:33.650 --> 00:30:37.489
<v James Loach>You get better context and it, you know,
you get the, I don't know if this is going

552
00:30:37.489 --> 00:30:41.670
<v James Loach>to be recorded, but I'm, I'm making hand
gestures like with a very steep slope or.

553
00:30:42.810 --> 00:30:47.020
<v James Loach>Like a flywheel gets going, you know, so
we're interested in, in making that happen,

554
00:30:47.099 --> 00:30:47.180
<v Liz McGinn>I

555
00:30:47.180 --> 00:30:47.459
<v James Loach>guess,

556
00:30:47.609 --> 00:30:48.189
<v Liz McGinn>longer term.

557
00:30:48.979 --> 00:30:52.739
<v Liz McGinn>Yeah, is it kind of coming about like
removing the human error for it, or do you

558
00:30:52.739 --> 00:30:57.780
<v Liz McGinn>think there might always kind of be a
human error based on the way that humans are

559
00:30:57.780 --> 00:30:59.319
<v Liz McGinn>kind of programming these.

560
00:31:00.569 --> 00:31:01.680
<v Liz McGinn>AI programs.

561
00:31:01.979 --> 00:31:02.530
<v Liz McGinn>Does that make sense?

562
00:31:03.540 --> 00:31:04.640
<v James Loach>Yeah, how do you mean?

563
00:31:05.300 --> 00:31:09.540
<v Liz McGinn>So like, um, I guess that's the, the one
thing about humans is like, obviously we

564
00:31:09.540 --> 00:31:13.300
<v Liz McGinn>have cognitive thinking and stuff like
that, but there's always this like room for

565
00:31:13.300 --> 00:31:14.060
<v Liz McGinn>human error.

566
00:31:14.920 --> 00:31:20.400
<v Liz McGinn>Would these AI robots kind of remove that
human error portion of it, or because

567
00:31:20.400 --> 00:31:24.880
<v Liz McGinn>humans are kind of building the program
itself, it might just kind of be programmed

568
00:31:24.880 --> 00:31:25.280
<v Liz McGinn>in.

569
00:31:28.140 --> 00:31:29.920
<v James Loach>Yeah, again, again, that's it's very
interesting.

570
00:31:30.699 --> 00:31:33.900
<v James Loach>Yeah, I mean, if, if models are just
trained how to maintain machines by reading

571
00:31:33.900 --> 00:31:37.900
<v James Loach>random stuff on the internet, you know,
there's some fuzziness to that, and, you

572
00:31:37.900 --> 00:31:42.219
<v James Loach>know, obviously mistakes could, could
happen, which is not what you want, right?

573
00:31:43.459 --> 00:31:49.119
<v James Loach>You want them to, yeah, be trained on
optimal material in this sort of clear, um.

574
00:31:50.579 --> 00:31:55.380
<v James Loach>Way Yeah, it is, it is, it's sort of
interesting, right?

575
00:31:55.569 --> 00:31:56.150
<v James Loach>so.

576
00:31:56.839 --> 00:32:00.540
<v James Loach>AI systems as we currently have, yeah,
that there is some, some error, error.

577
00:32:01.979 --> 00:32:05.319
<v James Loach>You know, error proneness there with
hallucinations and things like this.

578
00:32:05.660 --> 00:32:09.420
<v James Loach>Human beings also make their own kind of
errors of different sorts, you know.

579
00:32:09.579 --> 00:32:14.229
<v James Loach>Most human errors are to do with lack of
concentration or lack of listening to tasks

580
00:32:14.780 --> 00:32:15.910
<v James Loach>or things like this.

581
00:32:16.290 --> 00:32:19.500
<v James Loach>And so, you know, again, in the medium
term, you can imagine.

582
00:32:20.310 --> 00:32:23.349
<v James Loach>You know, humans and machines working
together quite well because you have sort of

583
00:32:23.349 --> 00:32:28.819
<v James Loach>different failure modes and some sort of
supportive, um, back, back and forth, um,

584
00:32:29.189 --> 00:32:32.579
<v James Loach>but you know, my long term view would be
that, that ultimately humans don't have a,

585
00:32:32.630 --> 00:32:34.650
<v James Loach>a massive amount to add there, right?

586
00:32:35.310 --> 00:32:39.329
<v James Loach>Um, but I, I, I think a nice analogy of
this is, is self-driving cars.

587
00:32:40.040 --> 00:32:45.000
<v James Loach>Yeah, you know, you're in a system now
where, as I understand it, something which,

588
00:32:45.079 --> 00:32:48.280
<v James Loach>you know, if you're Waymo or whatever, it
feels like it's a completely self-driving

589
00:32:48.280 --> 00:32:52.719
<v James Loach>car, but when you get into tricky
situations, there's somebody, you know, in the

590
00:32:52.719 --> 00:32:57.119
<v James Loach>background in a distant center that takes
over the car to steer it around this funny

591
00:32:57.119 --> 00:33:02.160
<v James Loach>thing, you know, and the system, the car
also can, you know, crash in like odd and

592
00:33:02.160 --> 00:33:04.119
<v James Loach>strange worrying ways and things like
this.

593
00:33:04.579 --> 00:33:06.959
<v James Loach>But I, I guess we, we know where this is
headed, right?

594
00:33:07.400 --> 00:33:10.810
<v James Loach>You know, either in, in 10 years' time or
whenever.

595
00:33:11.400 --> 00:33:13.280
<v James Loach>Yeah, not only will there be.

596
00:33:14.229 --> 00:33:17.540
<v James Loach>You know, safer than humans striving, but
will kind of some in some sense be

597
00:33:17.540 --> 00:33:18.719
<v James Loach>optimal, right?

598
00:33:18.829 --> 00:33:22.589
<v James Loach>So you're getting the absolute minimum
possible number of accidents, yeah, not just

599
00:33:22.589 --> 00:33:23.770
<v James Loach>better than humans.

600
00:33:24.310 --> 00:33:28.339
<v James Loach>And I, I, I can imagine this, this kind of
pattern just being ultimately repeated

601
00:33:28.339 --> 00:33:28.989
<v James Loach>everywhere.

602
00:33:29.709 --> 00:33:34.800
<v James Loach>Um, and like you said, not, not too soon,
um, and if, if things like that happen

603
00:33:34.800 --> 00:33:39.719
<v James Loach>very, very quickly, that causes a lot of
disturbance and craziness in the world and

604
00:33:39.719 --> 00:33:40.459
<v James Loach>all of that.

605
00:33:41.599 --> 00:33:47.719
<v James Loach>But, um, I really think with AI, you know,
what will happen will happen, right?

606
00:33:47.780 --> 00:33:52.010
<v James Loach>And, um, we should largely try to mitigate
the problems but largely enjoy the ride.

607
00:33:52.390 --> 00:33:52.410
<v James Loach>I think

608
00:33:52.410 --> 00:33:52.739
<v Liz McGinn>it's my,

609
00:33:53.489 --> 00:33:54.219
<v James Loach>my philosophy.

610
00:33:55.609 --> 00:33:56.030
<v Liz McGinn>Awesome.

611
00:33:56.329 --> 00:33:58.489
<v Liz McGinn>So I'm kind of uh up to my questions here.

612
00:33:58.569 --> 00:34:00.489
<v Liz McGinn>I have um kind of two more.

613
00:34:00.609 --> 00:34:05.290
<v Liz McGinn>So one, what is your favorite way that
you're using AI in your day to day?

614
00:34:05.530 --> 00:34:06.569
<v Liz McGinn>Just for funsies.

615
00:34:09.989 --> 00:34:16.870
<v James Loach>Um, yeah, so I think, um, using, using
agents to, to write software, um,

616
00:34:16.989 --> 00:34:18.830
<v James Loach>is, is, is quite nice.

617
00:34:19.199 --> 00:34:20.469
<v James Loach>And it's, it's, it's, um.

618
00:34:21.379 --> 00:34:24.010
<v James Loach>a sort of strange thing, yeah.

619
00:34:24.169 --> 00:34:27.729
<v James Loach>So, so you've, you've got the agent there
who, you know, for, for many projects

620
00:34:27.729 --> 00:34:32.290
<v James Loach>might, might really be writing most of the
actual code and you spend your time, um,

621
00:34:32.370 --> 00:34:36.449
<v James Loach>like directing it, you know, and checking
what it's done and things like that.

622
00:34:36.889 --> 00:34:40.129
<v James Loach>Um, but the reason that's interesting,
obviously, the thing that I just described in

623
00:34:40.129 --> 00:34:43.010
<v James Loach>itself is, is somewhat cool, but it's just
trying to think about how to do this

624
00:34:43.010 --> 00:34:43.790
<v James Loach>optimally.

625
00:34:44.580 --> 00:34:48.659
<v James Loach>Yeah, because you know you can't get into
a lot of wasting of time just trying to

626
00:34:48.659 --> 00:34:51.560
<v James Loach>get the system to do things where you
could, you could just be doing it yourself,

627
00:34:52.219 --> 00:34:58.250
<v James Loach>um, but yet the systems have potential to
do things, yeah, much faster and a greater

628
00:34:58.250 --> 00:34:59.379
<v James Loach>scale than you can, right?

629
00:34:59.699 --> 00:35:04.129
<v James Loach>So yeah, like, so yeah, in terms of
developing these things, so I think one example,

630
00:35:04.179 --> 00:35:07.899
<v James Loach>yeah, like coding agents, how to have like
kind of multiple coding, coding agents

631
00:35:07.899 --> 00:35:12.379
<v James Loach>working together, how to control them, and
how to kind of optimize that thing to.

632
00:35:12.830 --> 00:35:15.929
<v James Loach>Get best results and like actually really
do things quicker.

633
00:35:16.149 --> 00:35:16.629
<v Liz McGinn>Yeah.

634
00:35:17.189 --> 00:35:19.939
<v James Loach>um, and that, yeah, leads into the product
as well.

635
00:35:19.989 --> 00:35:24.909
<v James Loach>We are also playing similar games with,
with agents, uh, you know, to try and

636
00:35:24.909 --> 00:35:26.189
<v James Loach>improve many things we do.

637
00:35:26.629 --> 00:35:31.310
<v James Loach>But it, we feel it most visceral,
viscerally with our jobs, which is the actual

638
00:35:31.310 --> 00:35:32.760
<v James Loach>writing code and things, yeah.

639
00:35:33.250 --> 00:35:33.969
<v James Loach>um, so that's fun.

640
00:35:35.260 --> 00:35:35.899
<v Liz McGinn>Nice.

641
00:35:35.979 --> 00:35:36.419
<v Liz McGinn>OK.

642
00:35:36.699 --> 00:35:41.860
<v Liz McGinn>Um, and final question, what is kind of
something that you want everyone to know or

643
00:35:41.860 --> 00:35:46.939
<v Liz McGinn>any advice or um just kind of last
comments that we might have missed during the

644
00:35:46.939 --> 00:35:47.659
<v Liz McGinn>rest of this?

645
00:35:50.530 --> 00:35:54.840
<v James Loach>Yeah, I, I don't have any, any general
advice for people on this, this, I, I think,

646
00:35:54.889 --> 00:36:00.409
<v James Loach>I mean, people know that, you know, the
reality here that it's, um, sort of

647
00:36:00.409 --> 00:36:03.209
<v James Loach>exciting, interesting, and slightly
uncertain time.

648
00:36:04.100 --> 00:36:06.110
<v James Loach>Um, I don't know.

649
00:36:06.479 --> 00:36:07.719
<v James Loach>Yeah, I, I think.

650
00:36:08.489 --> 00:36:15.010
<v James Loach>A sensible take on all this stuff is to be
engaged with it and experiment, yeah, and

651
00:36:15.449 --> 00:36:15.810
<v James Loach>um.

652
00:36:16.959 --> 00:36:20.520
<v James Loach>Not be in denial of it and somehow have it
wash over you at some point in the

653
00:36:20.520 --> 00:36:21.100
<v James Loach>future.

654
00:36:21.719 --> 00:36:25.439
<v James Loach>So, you know, there's all kinds of hype
and there's all kinds of nonsense, but

655
00:36:25.439 --> 00:36:29.600
<v James Loach>there's something real there and I think,
yeah, we should all be, um, engaging with

656
00:36:29.600 --> 00:36:30.860
<v James Loach>it at a minimum, you know.

657
00:36:32.100 --> 00:36:34.110
<v James Loach>Yeah, not to, not to be taken by surprise.

658
00:36:36.189 --> 00:36:37.030
<v Liz McGinn>Understandable.

659
00:36:37.830 --> 00:36:38.270
<v Liz McGinn>Awesome.

660
00:36:38.520 --> 00:36:40.350
<v Liz McGinn>So thank you so much for joining me today.

661
00:36:40.429 --> 00:36:41.929
<v Liz McGinn>I really appreciate it.

662
00:36:42.949 --> 00:36:44.010
<v James Loach>Yeah, it's very nice to talk.

663
00:36:45.159 --> 00:36:49.520
<v Speaker 1>Please like and subscribe on all major
podcast channels and find out more about

664
00:36:49.520 --> 00:36:51.800
<v Speaker 1>Sensi Predictive maintenance at
Siemens.com.
