1
00:00:02,560 --> 00:00:05,359
Speaker 1: Weekend Breakfast the Future of.

2
00:00:05,200 --> 00:00:07,760
Speaker 2: Eight minutes after nine o'clock turn for us to talk

3
00:00:07,840 --> 00:00:10,799
Speaker 2: the future of and as always every second week, we

4
00:00:10,880 --> 00:00:16,240
Speaker 2: talk about artificial intelligence, where there's always new and exciting, interesting,

5
00:00:16,960 --> 00:00:20,760
Speaker 2: sometimes terrifying developments. And this week we're going to look

6
00:00:20,760 --> 00:00:25,640
Speaker 2: at two developments, two stories about AI in the news cycle.

7
00:00:25,720 --> 00:00:28,880
Speaker 2: The one is the Pentagon setting a deadline of yesterday

8
00:00:28,880 --> 00:00:32,680
Speaker 2: for Anthropic to abandon its ethics rules for its AI.

9
00:00:33,080 --> 00:00:39,519
Speaker 2: They threatened to cancel Pentagon work for Anthropic and even

10
00:00:39,760 --> 00:00:44,280
Speaker 2: designate them a supply chain risk if they didn't do

11
00:00:44,440 --> 00:00:48,920
Speaker 2: away with their rules, and so now we're asking the question,

12
00:00:49,040 --> 00:00:54,680
Speaker 2: is this the next frontier of AI in military operations?

13
00:00:54,720 --> 00:00:58,600
Speaker 2: Apparently it was used in the capture of Nicholas Muduro,

14
00:00:59,160 --> 00:01:02,960
Speaker 2: the Venezuela president, was kidnapped by the US. So joined

15
00:01:03,040 --> 00:01:05,759
Speaker 2: us always in studio by Chief Data and Analytics Office

16
00:01:05,760 --> 00:01:08,600
Speaker 2: aid F and B. He's also an author of two

17
00:01:08,760 --> 00:01:12,840
Speaker 2: must read books around AI. The first is African Artificial Intelligence,

18
00:01:13,160 --> 00:01:18,080
Speaker 2: the other is Sovereign Artificial Intelligence. Doctor Mark Nacilla joins

19
00:01:18,120 --> 00:01:21,160
Speaker 2: us in studio this morning. Doctor Nacilla, Good morning.

20
00:01:21,319 --> 00:01:22,360
Speaker 1: Good morning gooks.

21
00:01:22,640 --> 00:01:23,479
Speaker 2: It is another week.

22
00:01:23,920 --> 00:01:28,200
Speaker 1: Great to be here again. And the AI world still

23
00:01:28,280 --> 00:01:33,760
Speaker 1: very interesting. A lot happening, especially today's story about Anthropic

24
00:01:35,120 --> 00:01:37,759
Speaker 1: versus the United States.

25
00:01:38,000 --> 00:01:42,880
Speaker 2: So very interesting that this is well interesting but not

26
00:01:42,920 --> 00:01:46,040
Speaker 2: surprising that this is where we are. I think AI

27
00:01:46,360 --> 00:01:50,840
Speaker 2: was always perfectly positioned to be used in combat, in

28
00:01:50,920 --> 00:01:54,000
Speaker 2: military by you know, by armies around the world, by

29
00:01:54,280 --> 00:01:56,200
Speaker 2: navies around the world. It seemed like it was the

30
00:01:56,280 --> 00:01:59,360
Speaker 2: kind of thing that would be used in this way.

31
00:01:59,440 --> 00:02:01,640
Speaker 2: That's that's not surprising.

32
00:02:01,200 --> 00:02:06,080
Speaker 1: Right, It's not surprising in fact, when you look at

33
00:02:06,280 --> 00:02:10,160
Speaker 1: the amount of Spain into AI. The highest budget has

34
00:02:10,200 --> 00:02:16,440
Speaker 1: gone on defense and the use of automatic weapons. But

35
00:02:16,520 --> 00:02:19,639
Speaker 1: this story is very interesting because things have escalated so

36
00:02:19,760 --> 00:02:26,760
Speaker 1: fast this week beyond just the threat to terminate anthropics

37
00:02:27,160 --> 00:02:36,639
Speaker 1: contract with a Pentagon. Yesterday, Donald Trump burnt Anthropic from

38
00:02:36,720 --> 00:02:43,160
Speaker 1: using government systems, and immediately after that they strike a

39
00:02:43,240 --> 00:02:46,800
Speaker 1: new deal with open AI for the same purpose they

40
00:02:46,840 --> 00:02:48,760
Speaker 1: wanted to use Anthropic for.

41
00:02:49,000 --> 00:02:53,640
Speaker 2: So that's some ultimate open AI exactly right. So what

42
00:02:53,800 --> 00:02:57,320
Speaker 2: is the what are these I guess the ethic rules

43
00:02:57,480 --> 00:03:00,200
Speaker 2: that Anthropic had for AI. We spoke a little bit

44
00:03:00,200 --> 00:03:02,360
Speaker 2: about Anthropic the last time we met two weeks ago.

45
00:03:02,800 --> 00:03:09,320
Speaker 2: They're currently paying a Swedish researcher to teach their AI morals,

46
00:03:09,440 --> 00:03:13,560
Speaker 2: to teach the AI ethics. So hopefully yes, Amanda, Yes,

47
00:03:13,960 --> 00:03:16,760
Speaker 2: so it can be a force for good. Is that

48
00:03:16,840 --> 00:03:21,040
Speaker 2: part of where the fight isn't that Anthropic is they're

49
00:03:21,120 --> 00:03:24,000
Speaker 2: kind of asking these big questions, how do we make

50
00:03:24,080 --> 00:03:28,760
Speaker 2: this really powerful technology ethical and good? Which then makes

51
00:03:28,800 --> 00:03:31,600
Speaker 2: it difficult to use it for well, I guess for

52
00:03:31,760 --> 00:03:33,440
Speaker 2: bad necessarily.

53
00:03:33,240 --> 00:03:38,080
Speaker 1: Exactly, So before that you write Anthropicals in the news

54
00:03:38,120 --> 00:03:41,600
Speaker 1: for ethical AI reasons. Once they are appointed, Amanda a

55
00:03:41,680 --> 00:03:46,240
Speaker 1: scale to spear ahead the ethical AI strategy. But also

56
00:03:46,720 --> 00:03:50,160
Speaker 1: they were being spoken around being used by America for

57
00:03:50,240 --> 00:03:55,800
Speaker 1: surveillance in the capture of Maduro in Venezuela. But this

58
00:03:55,960 --> 00:04:02,440
Speaker 1: specific case, you know, the controversy came from when Anthropic

59
00:04:02,880 --> 00:04:08,840
Speaker 1: was reluctant to lift its safeguard rails. And this was

60
00:04:08,960 --> 00:04:14,760
Speaker 1: basically the that came from the defense sectory. Unless they

61
00:04:14,800 --> 00:04:19,120
Speaker 1: agreed to Trump's terms of use, their contract will be terminated.

62
00:04:20,320 --> 00:04:24,120
Speaker 1: And there's two major issues here that Anthropic had. One

63
00:04:24,240 --> 00:04:26,240
Speaker 1: they did not want their tools to be used for

64
00:04:27,080 --> 00:04:33,400
Speaker 1: mass domestic surveillance. The second parties the automated physical attacks

65
00:04:33,480 --> 00:04:37,680
Speaker 1: unless there was human supervision in those attacks. And this

66
00:04:37,720 --> 00:04:44,080
Speaker 1: is in summary, just the usage policy around responsible and

67
00:04:44,200 --> 00:04:51,600
Speaker 1: ethical aspects of their platforms. And basically Donald Trump invoked

68
00:04:52,080 --> 00:04:56,080
Speaker 1: the Defense Production Act and the Supply Chain Risk which

69
00:04:56,160 --> 00:04:59,080
Speaker 1: basically blacklists anthropic right.

70
00:05:00,040 --> 00:05:05,960
Speaker 2: And so these are interesting times in that so and

71
00:05:05,960 --> 00:05:09,000
Speaker 2: and I guess now that there's I mean, the deadline

72
00:05:09,040 --> 00:05:11,400
Speaker 2: was set for Friday, and Theropic was told to drop

73
00:05:11,440 --> 00:05:16,280
Speaker 2: its ethics rules, and then someone else folds that job, right,

74
00:05:16,279 --> 00:05:19,839
Speaker 2: because there are lots of companies, certainly in the US.

75
00:05:20,120 --> 00:05:25,640
Speaker 2: China also also has their options, right, And so are

76
00:05:25,640 --> 00:05:28,000
Speaker 2: we going to see I guess more of this where

77
00:05:28,240 --> 00:05:31,800
Speaker 2: maybe a company says I don't want my particular large

78
00:05:31,880 --> 00:05:35,320
Speaker 2: language model or agentic AI to be used in this way,

79
00:05:35,720 --> 00:05:38,240
Speaker 2: but someone else is going to say, where do I sign.

80
00:05:38,440 --> 00:05:41,360
Speaker 2: I'll you know, I'll take I'll take the steal, even

81
00:05:41,400 --> 00:05:44,720
Speaker 2: though for that company it was unethical or it was

82
00:05:44,760 --> 00:05:47,120
Speaker 2: a risk, or it was you know, not what they

83
00:05:47,160 --> 00:05:47,839
Speaker 2: wanted to do.

84
00:05:48,720 --> 00:05:53,800
Speaker 1: So this particular case is interesting because first of all,

85
00:05:54,760 --> 00:05:58,680
Speaker 1: Trump announced the burn an hour before the deadline. Oh

86
00:05:58,760 --> 00:06:03,279
Speaker 1: and we know very well Ultimate and Dario, but they

87
00:06:03,440 --> 00:06:05,680
Speaker 1: have are not in good terms. In fact, if you

88
00:06:05,800 --> 00:06:09,440
Speaker 1: saw at the AI submit in India, they refuse to

89
00:06:09,480 --> 00:06:12,880
Speaker 1: actually hold old hands, and so I don't want to

90
00:06:12,920 --> 00:06:16,600
Speaker 1: read so much into it, but it's a bit suspicious

91
00:06:16,680 --> 00:06:19,920
Speaker 1: because you know, announcing it an hour before and also

92
00:06:19,960 --> 00:06:23,920
Speaker 1: stry announcing this deal. And from what I understand is

93
00:06:24,240 --> 00:06:28,720
Speaker 1: that open AIS deal was on the same terms which

94
00:06:28,760 --> 00:06:34,320
Speaker 1: America was against, which is a bit suspicious. But there's

95
00:06:34,320 --> 00:06:36,880
Speaker 1: a couple of lessons here that I think from a

96
00:06:36,920 --> 00:06:42,880
Speaker 1: geopolitical intersect with national security and they you know, the

97
00:06:42,920 --> 00:06:47,600
Speaker 1: advancement of technology. The first one is that there's going

98
00:06:47,640 --> 00:06:50,560
Speaker 1: to be a bit of challenge between you know, corporate

99
00:06:50,839 --> 00:06:58,200
Speaker 1: or even AI startups versus states sovereignty, especially in protecting

100
00:06:58,360 --> 00:07:02,359
Speaker 1: national interests. If you have an AI startup or a

101
00:07:02,400 --> 00:07:07,200
Speaker 1: technology organization, what does this mean going forward when you

102
00:07:07,279 --> 00:07:10,880
Speaker 1: know an organization or country wants to leverage your technology

103
00:07:11,160 --> 00:07:17,080
Speaker 1: even against your principles. The second part is are we

104
00:07:17,360 --> 00:07:22,320
Speaker 1: entering an era of weaponizing company against companies that do

105
00:07:22,320 --> 00:07:26,400
Speaker 1: not want to go against our wills even though it's

106
00:07:26,400 --> 00:07:31,119
Speaker 1: an ethical so as I said, basically, what Donald Trump

107
00:07:31,120 --> 00:07:35,080
Speaker 1: has done is to blacklist, you know, by invoking the

108
00:07:35,120 --> 00:07:40,680
Speaker 1: supply and risk, you're declaring that anthropy, you're categorizing them

109
00:07:40,760 --> 00:07:46,880
Speaker 1: actually under adverse foreign country companies that I'm not about

110
00:07:46,880 --> 00:07:50,120
Speaker 1: to supply goods and services as well as information, and

111
00:07:50,200 --> 00:07:54,800
Speaker 1: therefore you're actually banning them against being used internally. So

112
00:07:55,120 --> 00:07:58,680
Speaker 1: Anthropic is not just supplying goods and services to America,

113
00:07:59,160 --> 00:08:03,840
Speaker 1: but they do actually supply other companies, like you know, Palentin.

114
00:08:05,520 --> 00:08:11,160
Speaker 1: The third one is, are we saying that your political

115
00:08:11,240 --> 00:08:15,160
Speaker 1: pressure must end safety because basically this is what has happened.

116
00:08:15,720 --> 00:08:19,520
Speaker 1: I do compliment Anthropic for you know, holding up, but

117
00:08:20,360 --> 00:08:22,880
Speaker 1: the pressure has resulted to them being kick aside, and

118
00:08:22,960 --> 00:08:27,040
Speaker 1: there's another alternative organization that's actually doing the same thing,

119
00:08:27,400 --> 00:08:31,360
Speaker 1: which was against And the fourth one is the technical

120
00:08:31,440 --> 00:08:35,520
Speaker 1: readiness of these tools doesn't matter anymore. From what I understand,

121
00:08:35,600 --> 00:08:40,160
Speaker 1: the sea of Anthropic told the US government while these

122
00:08:40,200 --> 00:08:43,320
Speaker 1: tools were very good, there was still a dangerous gap

123
00:08:43,400 --> 00:08:49,960
Speaker 1: between what these tools could do versus the military desire. Now,

124
00:08:50,200 --> 00:08:53,200
Speaker 1: even what we're saying is that America is willing to

125
00:08:53,280 --> 00:08:56,040
Speaker 1: actually use these technologies even though they are not as

126
00:08:56,200 --> 00:09:01,840
Speaker 1: accurate as they are to launch attacks also make you

127
00:09:01,840 --> 00:09:05,920
Speaker 1: know their citizens for their own you know, strategies.

128
00:09:06,320 --> 00:09:09,280
Speaker 2: Yes, and so I guess I mean right at the

129
00:09:09,280 --> 00:09:12,439
Speaker 2: beginning of our conversations, right I mean when we started

130
00:09:12,840 --> 00:09:15,400
Speaker 2: this feature, one of the things you said was we

131
00:09:15,480 --> 00:09:20,439
Speaker 2: need to be thinking about who or what organizations end

132
00:09:20,559 --> 00:09:24,960
Speaker 2: up with these really powerful technologies, ideally in an ideal world,

133
00:09:25,000 --> 00:09:27,480
Speaker 2: and unfortunately we don't live in an ideal world. You

134
00:09:27,600 --> 00:09:30,439
Speaker 2: want to have these technologies in the hands of people

135
00:09:30,840 --> 00:09:34,520
Speaker 2: who mean to use them for good, so improving literacy,

136
00:09:35,080 --> 00:09:38,520
Speaker 2: access to healthcare, you know, all of those I guess things.

137
00:09:39,160 --> 00:09:42,600
Speaker 2: But we are seeing though that And this was the

138
00:09:42,600 --> 00:09:45,600
Speaker 2: thing you raise that it's unlikely that and I put

139
00:09:45,640 --> 00:09:48,640
Speaker 2: this in in d commas. Good guys get to the tech,

140
00:09:49,000 --> 00:09:51,000
Speaker 2: the guys with the money, the guys who want to

141
00:09:51,080 --> 00:09:54,000
Speaker 2: perhaps fight more efficient war, they are the people that

142
00:09:54,040 --> 00:09:56,560
Speaker 2: are going to get to it first, exactly. And is

143
00:09:56,600 --> 00:09:57,439
Speaker 2: this what you meant?

144
00:09:57,760 --> 00:10:01,199
Speaker 1: This is what I meant, right, including now the most

145
00:10:01,240 --> 00:10:04,760
Speaker 1: powerful people can get to it, whether you want them

146
00:10:04,800 --> 00:10:05,520
Speaker 1: to have it or not.

147
00:10:05,920 --> 00:10:09,880
Speaker 2: Right, Yes, So now that we are here, I mean,

148
00:10:09,920 --> 00:10:13,520
Speaker 2: the horse has bolted. We now do have governments and

149
00:10:13,760 --> 00:10:16,280
Speaker 2: arguably one of the biggest militaries in the world now

150
00:10:16,320 --> 00:10:19,960
Speaker 2: saying we're going to make use of really sophisticated agentic

151
00:10:21,440 --> 00:10:25,040
Speaker 2: AI as part of our operations. What does that mean

152
00:10:25,200 --> 00:10:28,959
Speaker 2: for the rest of us. Should we be concerned, should

153
00:10:28,960 --> 00:10:30,200
Speaker 2: we be worried?

154
00:10:31,480 --> 00:10:36,480
Speaker 1: We should be concerned for the four reasons I've just stated,

155
00:10:36,679 --> 00:10:41,199
Speaker 1: you know, around you know, the intersection between steps sovereignty

156
00:10:41,360 --> 00:10:45,120
Speaker 1: as well as technology and and and just your politics.

157
00:10:45,600 --> 00:10:49,120
Speaker 1: We should be concerned that going forward, you know, powerful

158
00:10:49,240 --> 00:10:55,520
Speaker 1: companies will weaponize their will againist organizations that are you know,

159
00:10:55,640 --> 00:10:58,560
Speaker 1: resistant to their wishes if we thought they're an ethical

160
00:10:59,360 --> 00:11:03,280
Speaker 1: we should be concerned that, you know, geopolitical pressure will

161
00:11:03,360 --> 00:11:07,080
Speaker 1: has is actually ending safety and those who can access

162
00:11:07,120 --> 00:11:08,120
Speaker 1: technology will use it.

163
00:11:08,720 --> 00:11:09,520
Speaker 2: But I think.

164
00:11:10,880 --> 00:11:15,559
Speaker 1: We we talked about, you know, we need to rethink

165
00:11:15,640 --> 00:11:22,200
Speaker 1: our strategy and not trash into developing these technologies, ensuring

166
00:11:22,280 --> 00:11:27,040
Speaker 1: that we've got you know, responsible and ethical AI embedded

167
00:11:27,080 --> 00:11:32,040
Speaker 1: in the way we operate a society. Because what America

168
00:11:32,160 --> 00:11:35,480
Speaker 1: is doing, it wants to have access to a capability,

169
00:11:35,559 --> 00:11:39,400
Speaker 1: because that capability can perform some functions. There's there are

170
00:11:39,440 --> 00:11:41,120
Speaker 1: two functions, and.

171
00:11:41,000 --> 00:11:43,679
Speaker 3: It's it's it's it's it's it's it's it's made it

172
00:11:43,760 --> 00:11:46,679
Speaker 3: clear that it's it's gonna get access to that technology,

173
00:11:46,720 --> 00:11:52,160
Speaker 3: whether anthropic ones or not, whether open AI once or not. Uh.

174
00:11:52,200 --> 00:11:55,679
Speaker 1: And there's a couple of things that we must think

175
00:11:55,720 --> 00:11:59,360
Speaker 1: from a society perspective as well as all organizations that

176
00:11:59,440 --> 00:12:02,800
Speaker 1: I living technology, and they range from make sure there's

177
00:12:02,840 --> 00:12:06,280
Speaker 1: a clear alignment on what the use of this technology

178
00:12:06,320 --> 00:12:09,640
Speaker 1: is meant for, right, because if there's no clear alignment,

179
00:12:09,840 --> 00:12:12,400
Speaker 1: then you'll end up being in a position where we're

180
00:12:12,440 --> 00:12:16,440
Speaker 1: seeing now. America feels like when it has certain use

181
00:12:16,480 --> 00:12:19,040
Speaker 1: of this need to use the technology, then it must

182
00:12:19,040 --> 00:12:21,160
Speaker 1: actually make sure it gets access to it.

183
00:12:21,360 --> 00:12:21,560
Speaker 3: Yeah.

184
00:12:21,640 --> 00:12:24,680
Speaker 2: Yeah, and so I mean in history, we've seen various

185
00:12:24,800 --> 00:12:27,800
Speaker 2: kind of races. So there was a time where America

186
00:12:28,000 --> 00:12:30,640
Speaker 2: was in a space race with the USSR to try

187
00:12:30,640 --> 00:12:32,559
Speaker 2: to get the first man in space, the first man

188
00:12:32,600 --> 00:12:35,480
Speaker 2: on the moon, et cetera. Then we saw an arms

189
00:12:35,559 --> 00:12:38,040
Speaker 2: race where countries were trying to be the ones who

190
00:12:38,080 --> 00:12:41,520
Speaker 2: had the biggest amount of armaments. It feels as though

191
00:12:42,120 --> 00:12:45,760
Speaker 2: with this kind of development in AI that we potentially

192
00:12:45,840 --> 00:12:48,640
Speaker 2: will have another arms race, but it will be about

193
00:12:48,800 --> 00:12:53,800
Speaker 2: who's able to deploy this technology first or tested in

194
00:12:53,880 --> 00:12:55,960
Speaker 2: the field. Who's going to be able to put this

195
00:12:56,000 --> 00:13:00,120
Speaker 2: technology in maybe the human or robots that alibab I

196
00:13:00,160 --> 00:13:02,080
Speaker 2: was talking about a couple of weeks ago. You know this,

197
00:13:02,200 --> 00:13:06,080
Speaker 2: how do we use physical AI in conflict. It feels

198
00:13:06,080 --> 00:13:08,679
Speaker 2: as though this could be the next and then there

199
00:13:08,720 --> 00:13:10,040
Speaker 2: was a time where there was you know, in arms

200
00:13:10,120 --> 00:13:12,720
Speaker 2: race relating to the nuclear bomb. Right, it feels like

201
00:13:12,760 --> 00:13:16,760
Speaker 2: this is going to be the next frontier of defense

202
00:13:16,920 --> 00:13:20,920
Speaker 2: and war and countries trying to protect their sovereignty. Is

203
00:13:20,960 --> 00:13:26,680
Speaker 2: who has deployed or has tested AI to defend itself

204
00:13:26,720 --> 00:13:29,840
Speaker 2: whatever that looks like, or to boost up or bolster

205
00:13:30,040 --> 00:13:30,640
Speaker 2: its army.

206
00:13:31,080 --> 00:13:34,040
Speaker 1: That feels like that could be where we go. Yes,

207
00:13:34,720 --> 00:13:38,839
Speaker 1: and this is not surprising. America has what we call

208
00:13:38,920 --> 00:13:42,240
Speaker 1: the Defense Production Act. This is a nineteen fifty cold

209
00:13:42,400 --> 00:13:46,200
Speaker 1: or RA Act that allows the US president to direct

210
00:13:46,320 --> 00:13:51,240
Speaker 1: any industry organization to prioritize defense needs despite of any

211
00:13:51,240 --> 00:13:57,559
Speaker 1: ethical considerations. And it's used against the Korean War and

212
00:13:58,040 --> 00:14:03,160
Speaker 1: blacklisting or the designating Anthropic as a supply change risk

213
00:14:03,320 --> 00:14:07,959
Speaker 1: is basically weaponizing the intention to use the technology and

214
00:14:08,040 --> 00:14:11,560
Speaker 1: declaring the most against American will. Right, So this is

215
00:14:11,800 --> 00:14:15,680
Speaker 1: almost an extension of human decisions, if you may think

216
00:14:15,720 --> 00:14:19,600
Speaker 1: about it. Now, we have a technology and the government

217
00:14:19,640 --> 00:14:22,800
Speaker 1: will do whatever it can to use that technology for

218
00:14:23,200 --> 00:14:24,320
Speaker 1: its intention.

219
00:14:25,000 --> 00:14:28,400
Speaker 2: Yes, and so then is this could then this be?

220
00:14:29,000 --> 00:14:32,320
Speaker 2: Could this then be the kind of template or the

221
00:14:32,440 --> 00:14:36,480
Speaker 2: kind of thing we see more of where a government

222
00:14:36,800 --> 00:14:39,240
Speaker 2: and entities is we want to use this technology for

223
00:14:39,520 --> 00:14:42,760
Speaker 2: X a particular country, a particular company says, we don't

224
00:14:42,880 --> 00:14:45,400
Speaker 2: do that. That goes against our roles, and we just

225
00:14:45,480 --> 00:14:50,320
Speaker 2: find someone else. They will always be someone else exactly.

226
00:14:50,360 --> 00:14:55,920
Speaker 1: So what this sets up, presidents, is that whether you

227
00:14:56,120 --> 00:14:59,240
Speaker 1: like it or not, we will leverate the same services

228
00:14:59,240 --> 00:15:02,600
Speaker 1: from you or not. And in my view, maybe it's

229
00:15:02,640 --> 00:15:06,280
Speaker 1: a big lesson to technology companies. Do not just develop

230
00:15:06,360 --> 00:15:10,920
Speaker 1: technology that could do anything. Develop specialized technologies. If you

231
00:15:11,000 --> 00:15:15,440
Speaker 1: develop a technology that could perform mass surveillance, then it

232
00:15:15,480 --> 00:15:18,360
Speaker 1: could end up being used for that. If you develop

233
00:15:18,400 --> 00:15:22,240
Speaker 1: a technology that could be used for autonomous whip on attacks,

234
00:15:22,360 --> 00:15:27,120
Speaker 1: then it could be used for that. Then limit those capabilities.

235
00:15:27,680 --> 00:15:30,680
Speaker 1: I think a hard one will be maybe adapt the

236
00:15:30,840 --> 00:15:35,400
Speaker 1: human in the loop so that one of the issues

237
00:15:35,520 --> 00:15:38,440
Speaker 1: Anthropic hard It says, it's technology is not designed to

238
00:15:38,520 --> 00:15:41,360
Speaker 1: have someone to provide oversight when you launch an attack,

239
00:15:42,160 --> 00:15:45,400
Speaker 1: But then build it upfront so that when it's being used,

240
00:15:45,800 --> 00:15:50,400
Speaker 1: you know, the American defense can make sure they put

241
00:15:50,480 --> 00:15:53,040
Speaker 1: people behind it other than use it the way it is.

242
00:15:53,920 --> 00:15:57,040
Speaker 1: And then just when you look at the aftermath of

243
00:15:57,120 --> 00:16:01,160
Speaker 1: what will now happen with the anthropic. It's important that

244
00:16:01,280 --> 00:16:05,680
Speaker 1: transparency is prioritized on how these tools are, you know,

245
00:16:06,160 --> 00:16:12,440
Speaker 1: spearheading responsible application. Make sure you provide audits of how

246
00:16:13,320 --> 00:16:17,200
Speaker 1: they're ensuring safety and use across the world.

247
00:16:17,360 --> 00:16:21,640
Speaker 2: Yeah, yes, received, I think a timely message. It kind

248
00:16:21,640 --> 00:16:24,800
Speaker 2: of leads into our second part of the conversation, and

249
00:16:25,120 --> 00:16:28,240
Speaker 2: it is about the impact that AI could have on us,

250
00:16:28,240 --> 00:16:32,480
Speaker 2: and it says AI breeding human thinking is out of

251
00:16:32,480 --> 00:16:34,840
Speaker 2: the question. Just like before the cell phone, I could

252
00:16:34,920 --> 00:16:39,480
Speaker 2: remember twenty telephone numbers. Now I can't remember one contact number.

253
00:16:39,880 --> 00:16:44,600
Speaker 2: AI will destroy the biological cognitive horsepower of the human brain.

254
00:16:44,720 --> 00:16:47,560
Speaker 2: Humans will be brain dead. And there was a peace

255
00:16:47,600 --> 00:16:50,320
Speaker 2: in the Guardian. It was actually a letter someone wrote

256
00:16:50,360 --> 00:16:52,560
Speaker 2: in and said, I'm worried my brother. My boyfriend's use

257
00:16:52,600 --> 00:16:56,120
Speaker 2: of AI is affecting his ability to think for himself.

258
00:16:56,160 --> 00:16:58,880
Speaker 2: And it says, my boyfriend of eight years was forty four.

259
00:16:58,920 --> 00:17:01,920
Speaker 2: He ADHD. He runs his own business. He's always struggled

260
00:17:01,920 --> 00:17:06,120
Speaker 2: with admin and mundane tasks, but AI has revolutionized how

261
00:17:06,200 --> 00:17:08,399
Speaker 2: he works. I'm not worried. You can't seem to do

262
00:17:08,440 --> 00:17:12,199
Speaker 2: anything without AI. He's a heavy chat GPT user and

263
00:17:12,400 --> 00:17:16,639
Speaker 2: uses it even if there are better, non AI alternatives available.

264
00:17:16,880 --> 00:17:19,720
Speaker 2: So for instance, he'll ask for the train times rather

265
00:17:19,760 --> 00:17:22,680
Speaker 2: than using train line, which is an app, even though

266
00:17:22,720 --> 00:17:26,520
Speaker 2: it is less accurate. He just got his chat GPT wrapped.

267
00:17:26,800 --> 00:17:30,679
Speaker 2: He's in the top zero point three percent of users worldwide.

268
00:17:30,760 --> 00:17:33,760
Speaker 2: I worry about his ability to think independently, as well

269
00:17:33,800 --> 00:17:36,200
Speaker 2: as the environmental impact. I know it's a useful talk

270
00:17:36,200 --> 00:17:39,480
Speaker 2: for him, but he uses it for everything in life.

271
00:17:39,720 --> 00:17:42,840
Speaker 2: And so a concern here about what happens when someone

272
00:17:42,920 --> 00:17:47,720
Speaker 2: I guess becomes quite reliant, almost over reliant, on a technology,

273
00:17:48,160 --> 00:17:52,159
Speaker 2: particularly something like AI. We also know this is happening

274
00:17:52,160 --> 00:17:55,880
Speaker 2: at a time where there's this quite significant decline in literacy.

275
00:17:56,359 --> 00:18:00,679
Speaker 2: So just across the board, young adults, teenagers just not

276
00:18:00,960 --> 00:18:06,240
Speaker 2: able to read for understanding, read for meaning like their

277
00:18:06,400 --> 00:18:09,480
Speaker 2: parents and the generations before them, And so how do

278
00:18:09,560 --> 00:18:11,600
Speaker 2: we make sense of that? That's kind of like over

279
00:18:11,720 --> 00:18:14,240
Speaker 2: reliance on technology.

280
00:18:15,320 --> 00:18:17,760
Speaker 1: It reminds me of what we call the Google effects,

281
00:18:17,760 --> 00:18:22,760
Speaker 1: where all of a sudden people started forgetting things because

282
00:18:22,840 --> 00:18:27,480
Speaker 1: they knew Google Google. Remember, but this is a serious problem,

283
00:18:27,560 --> 00:18:32,040
Speaker 1: and maybe and and and and and and underlooked area

284
00:18:32,119 --> 00:18:36,399
Speaker 1: because we're much more focused on the value or the

285
00:18:36,440 --> 00:18:41,440
Speaker 1: profit that will come from technology U, especially in preserving

286
00:18:41,520 --> 00:18:45,240
Speaker 1: human intelligence. And and there's two parts of the of

287
00:18:45,280 --> 00:18:50,320
Speaker 1: the one the impact on digiorating cognitive skills and abilities,

288
00:18:50,960 --> 00:18:54,359
Speaker 1: but also the vulnerability of people who have got conditions.

289
00:18:54,880 --> 00:18:59,360
Speaker 1: So I think in this case, the lady's postriend has

290
00:18:59,359 --> 00:19:07,120
Speaker 1: got attention deficit disorder, Yes, and this means that they're

291
00:19:07,240 --> 00:19:09,680
Speaker 1: likely to create over dependence on these technology that I

292
00:19:09,680 --> 00:19:12,720
Speaker 1: remember a few weeks ago we talked about AI psychosis. Yes,

293
00:19:12,760 --> 00:19:16,280
Speaker 1: and people who are falling victim through this addiction are

294
00:19:16,320 --> 00:19:22,040
Speaker 1: the ones actually uh with psychological issues or mental disorders

295
00:19:22,400 --> 00:19:25,879
Speaker 1: and they're finding answers in them. The impact here and

296
00:19:25,960 --> 00:19:29,520
Speaker 1: even to the normal people is one we're seeing obviously,

297
00:19:29,720 --> 00:19:35,480
Speaker 1: the decline in you know, human intelligence skills like critical thinking, uh,

298
00:19:35,800 --> 00:19:39,000
Speaker 1: you know, the problem solving now. In fact, studies are

299
00:19:39,040 --> 00:19:43,640
Speaker 1: showing that, you know, the neural composition of people are

300
00:19:43,680 --> 00:19:48,000
Speaker 1: not using these tools is much intact than those who

301
00:19:48,000 --> 00:19:52,360
Speaker 1: are using them. Then talked about you know, the impaired memory.

302
00:19:52,400 --> 00:19:56,359
Speaker 1: People are so much dependent on technology they forget quickly.

303
00:19:56,880 --> 00:20:01,920
Speaker 1: The shortened attention lifespans, so you're not able to focus

304
00:20:01,960 --> 00:20:06,600
Speaker 1: for a long time, and then a very concerning one

305
00:20:06,720 --> 00:20:09,760
Speaker 1: is what we call the cognitive in Nasha, there's a

306
00:20:09,760 --> 00:20:14,479
Speaker 1: bit of people are becoming reluctant to initiate independent thoughts

307
00:20:15,119 --> 00:20:21,000
Speaker 1: and express yourself independently. The creativity is going down and

308
00:20:21,119 --> 00:20:25,360
Speaker 1: just inferiority complex and I think this is why this

309
00:20:25,480 --> 00:20:31,040
Speaker 1: lady has those concerns. And I think beyond just guardrails,

310
00:20:32,200 --> 00:20:35,920
Speaker 1: the technology world and even organizations in the human capital

311
00:20:35,960 --> 00:20:40,040
Speaker 1: areas must rethink what it means risk killing and the

312
00:20:40,080 --> 00:20:43,720
Speaker 1: future of work, the future of expressing yourself, the limitations

313
00:20:44,160 --> 00:20:47,359
Speaker 1: of what technology must do for people so that we

314
00:20:47,440 --> 00:20:51,080
Speaker 1: do not lose the human side of the intelligence.

315
00:20:51,560 --> 00:20:55,720
Speaker 2: It's interesting that the Internet and it's kind of immense power,

316
00:20:56,160 --> 00:20:59,200
Speaker 2: its ability to kind of give us access to information.

317
00:20:59,440 --> 00:21:01,720
Speaker 2: I mean you mean in the Google effect, where you know,

318
00:21:01,720 --> 00:21:05,040
Speaker 2: you kind of ask someone, you know, who was the

319
00:21:05,080 --> 00:21:07,920
Speaker 2: second president of South Africa instead of kind of giving

320
00:21:07,960 --> 00:21:10,040
Speaker 2: it a bit of thought because thoughts thinking takes a

321
00:21:10,040 --> 00:21:12,239
Speaker 2: bit of time. You go, okay, so there was now

322
00:21:12,280 --> 00:21:14,080
Speaker 2: someone Della and to clerk and then Tabo was the

323
00:21:14,119 --> 00:21:17,200
Speaker 2: second deputy. Ah, it was table Maggie. Actually you kind

324
00:21:17,200 --> 00:21:19,040
Speaker 2: of google it because it's there. Google is there it

325
00:21:19,119 --> 00:21:23,119
Speaker 2: will answer my question. And it's interesting that technology, and

326
00:21:23,160 --> 00:21:26,560
Speaker 2: it's in the in the many ways that it's been transformative,

327
00:21:26,960 --> 00:21:30,040
Speaker 2: has also in many ways kind of made us need it.

328
00:21:30,080 --> 00:21:35,040
Speaker 2: So Facebook suddenly you could access friends, family, colleagues all

329
00:21:35,080 --> 00:21:37,240
Speaker 2: over the world, and then people got stuck on that

330
00:21:38,000 --> 00:21:41,280
Speaker 2: rolling timeline. That's rolling timeline, right, because it gains exactly

331
00:21:41,280 --> 00:21:44,040
Speaker 2: what I wanted exactly. And so it is interesting that

332
00:21:44,119 --> 00:21:48,320
Speaker 2: with all technologies there's almost like this warning that, yes,

333
00:21:48,600 --> 00:21:53,000
Speaker 2: use it, use it to do important work, work you

334
00:21:53,040 --> 00:21:56,640
Speaker 2: can't do on your own, but don't get lost in it.

335
00:21:56,880 --> 00:21:59,679
Speaker 2: And it feels as though we don't quite learn that

336
00:21:59,760 --> 00:22:03,240
Speaker 2: lets with any technology. I think AI is just the

337
00:22:03,359 --> 00:22:09,119
Speaker 2: latest iteration, you know, our cell phone screens. How we

338
00:22:09,280 --> 00:22:12,080
Speaker 2: just once there's a new technology, we kind of hand

339
00:22:12,200 --> 00:22:16,080
Speaker 2: over quite a lot of our thinking and autonomy over

340
00:22:16,160 --> 00:22:17,400
Speaker 2: to it, exactly.

341
00:22:17,440 --> 00:22:20,320
Speaker 1: And we've all been in the situation where someone told

342
00:22:20,359 --> 00:22:24,199
Speaker 1: you I'm going to think about it. You've promised your boss,

343
00:22:24,440 --> 00:22:27,400
Speaker 1: your partner, your friends, I'm going to think about it.

344
00:22:28,160 --> 00:22:33,080
Speaker 1: That statement is actually it is actually evolving, and in

345
00:22:33,160 --> 00:22:38,080
Speaker 1: a lot of organizations, especially leaders, they've initiated processes to

346
00:22:38,160 --> 00:22:41,400
Speaker 1: make sure that thinking happens so that these decisions are

347
00:22:41,440 --> 00:22:47,800
Speaker 1: made leveraging cognitive skills. In this case, there's bit concern

348
00:22:47,920 --> 00:22:51,679
Speaker 1: now the thinking process is actually being eroded away and

349
00:22:51,720 --> 00:22:55,600
Speaker 1: people are relying on technology. There's two things here that

350
00:22:55,800 --> 00:22:59,320
Speaker 1: actually I would recommend that must be done for organizations,

351
00:22:59,400 --> 00:23:03,520
Speaker 1: partners and individuals. The first one is to identify emotional

352
00:23:03,720 --> 00:23:09,040
Speaker 1: reasons behind over dependence on technology. And if we don't

353
00:23:09,080 --> 00:23:12,120
Speaker 1: do that, you're not able to understand what is actually

354
00:23:12,200 --> 00:23:14,720
Speaker 1: driving the overdependence that addiction.

355
00:23:15,600 --> 00:23:19,440
Speaker 2: So what is the what For instance, you know many people.

356
00:23:19,280 --> 00:23:25,879
Speaker 1: Are lonely, lonely and excitement, you know something is making

357
00:23:25,920 --> 00:23:30,680
Speaker 1: them much comfortable. The second part is just encourage people

358
00:23:30,680 --> 00:23:32,760
Speaker 1: to go back to all the way of doing things

359
00:23:33,160 --> 00:23:38,159
Speaker 1: that make them involved as human beings. I don't know

360
00:23:38,160 --> 00:23:41,520
Speaker 1: if you've had about the concept of you know, learning

361
00:23:41,560 --> 00:23:44,560
Speaker 1: from the future of the or learning from you know,

362
00:23:45,359 --> 00:23:51,320
Speaker 1: young leaders. So it's almost reverse mentorship where because you're

363
00:23:51,359 --> 00:23:55,199
Speaker 1: a leader, sometimes spend time with young people and instead

364
00:23:55,200 --> 00:23:59,240
Speaker 1: of mentoring them, let them mentor you so that you

365
00:23:59,280 --> 00:24:02,280
Speaker 1: can human eye is how you learn from them, but

366
00:24:04,080 --> 00:24:09,560
Speaker 1: also clearly defined tasks and things that people must do

367
00:24:09,640 --> 00:24:13,800
Speaker 1: as human beings and draw the line and drive that accountability.

368
00:24:13,880 --> 00:24:17,200
Speaker 2: Yes, and so again even with that, the task that

369
00:24:17,320 --> 00:24:21,520
Speaker 2: humans or tasked to be done by human beings, I mean,

370
00:24:21,800 --> 00:24:24,320
Speaker 2: what does that mean? Only because I think of so

371
00:24:24,400 --> 00:24:27,960
Speaker 2: many things that can now be done, either through automation

372
00:24:28,720 --> 00:24:31,680
Speaker 2: or through agentic AI itself. There so many things which

373
00:24:31,720 --> 00:24:34,320
Speaker 2: we thought were things only we could do. Right, we

374
00:24:34,359 --> 00:24:36,840
Speaker 2: thought we were special. We could think we could create culture,

375
00:24:37,200 --> 00:24:40,240
Speaker 2: we could build relationships. And now with AIS like Claude,

376
00:24:40,280 --> 00:24:45,320
Speaker 2: for instance, Claude can be empathetic. I thought Claude can

377
00:24:45,359 --> 00:24:47,399
Speaker 2: see that you are tired or will not see, but

378
00:24:47,720 --> 00:24:51,720
Speaker 2: you know. And so even that, like, what are those

379
00:24:51,760 --> 00:24:55,960
Speaker 2: things that humans can do? Are there still things that

380
00:24:56,000 --> 00:25:00,359
Speaker 2: only us as a species can still do? The AI

381
00:25:00,480 --> 00:25:01,360
Speaker 2: cant for instance.

382
00:25:02,160 --> 00:25:06,719
Speaker 1: So from a task simulation perspective, while engaging on those platforms,

383
00:25:06,720 --> 00:25:11,359
Speaker 1: you would actually feel like this technology can you know,

384
00:25:11,480 --> 00:25:16,080
Speaker 1: execute human functions. But I think that sometimes can be misleading.

385
00:25:16,840 --> 00:25:20,000
Speaker 1: When you're interacting with the physical world and people, you

386
00:25:20,720 --> 00:25:24,919
Speaker 1: realize that the way we express ourselves as human beings

387
00:25:25,040 --> 00:25:28,240
Speaker 1: is still unique and the technology is limited, and therefore

388
00:25:28,480 --> 00:25:33,320
Speaker 1: we must caution. Even though you interact with Cloud or Gemini,

389
00:25:33,400 --> 00:25:35,600
Speaker 1: you feel like, oh, you know, this thing is human

390
00:25:35,600 --> 00:25:38,480
Speaker 1: and it's providing that human touch in a real world environment.

391
00:25:38,480 --> 00:25:43,040
Speaker 1: There is limitations, and I think regulators should condition organizations

392
00:25:43,160 --> 00:25:47,399
Speaker 1: not to underlook the role of human judgment, human expression

393
00:25:47,480 --> 00:25:50,399
Speaker 1: so that we can maintain that cognitive skill in the

394
00:25:50,440 --> 00:25:55,520
Speaker 1: way we make decisions, but also just monitor and stress

395
00:25:55,560 --> 00:25:58,760
Speaker 1: test this on an ongoing basis so that people are

396
00:25:58,760 --> 00:26:01,159
Speaker 1: not over depending. You know, there's a big risk of

397
00:26:01,240 --> 00:26:04,679
Speaker 1: what to call decision fatigue where people just accept what

398
00:26:04,760 --> 00:26:08,359
Speaker 1: technology tells you without thinking. So you need to test

399
00:26:08,640 --> 00:26:13,879
Speaker 1: that and maybe reward you know, critical thinking or you know,

400
00:26:13,960 --> 00:26:18,080
Speaker 1: cognitive application of skills so that we don't use the

401
00:26:18,160 --> 00:26:21,760
Speaker 1: human aspect or else we you know, that's one easy

402
00:26:21,800 --> 00:26:25,040
Speaker 1: way of becoming becoming a digital colony where you create

403
00:26:25,200 --> 00:26:29,120
Speaker 1: so much dependency that life is now defined by how

404
00:26:29,160 --> 00:26:30,600
Speaker 1: you interact with the technology.

405
00:26:30,880 --> 00:26:34,560
Speaker 2: Right, Yes, and so that is the risk with agentic AI.

406
00:26:35,119 --> 00:26:39,679
Speaker 1: That's what happens that comes with all these technologies agentic ai,

407
00:26:39,920 --> 00:26:42,840
Speaker 1: all platforms where they become a goal to think for.

408
00:26:43,240 --> 00:26:46,280
Speaker 1: You know, they become a tool that advises you on

409
00:26:46,440 --> 00:26:49,440
Speaker 1: a relationship or a decision in life, a tool that

410
00:26:49,800 --> 00:26:53,120
Speaker 1: you're trying to use it to strategize with the business, right,

411
00:26:53,160 --> 00:26:55,760
Speaker 1: you know, and and basically if you find yourself doing that,

412
00:26:55,840 --> 00:26:57,160
Speaker 1: then you actually don't have a life.

413
00:26:57,600 --> 00:27:01,119
Speaker 2: Oh yes, So what does have a life in this

414
00:27:01,320 --> 00:27:05,000
Speaker 2: age of technology? Is like one of our questions that

415
00:27:05,080 --> 00:27:08,600
Speaker 2: we need to be thinking about outside of you know,

416
00:27:08,840 --> 00:27:12,240
Speaker 2: my relationship with Claude in Gemini and chatipt What does

417
00:27:12,359 --> 00:27:15,480
Speaker 2: my life in the real world in a taxile way?

418
00:27:15,560 --> 00:27:16,399
Speaker 2: What does that look like?

419
00:27:17,240 --> 00:27:22,280
Speaker 1: So being involved defines actually your life, being involved with

420
00:27:22,320 --> 00:27:27,880
Speaker 1: your decisions and the impact or outcomes of your decisions

421
00:27:27,880 --> 00:27:30,000
Speaker 1: and the next one and the next one. But if

422
00:27:30,160 --> 00:27:33,919
Speaker 1: technology is doing that for you, then it's actually running

423
00:27:33,960 --> 00:27:38,199
Speaker 1: your life, and you're just confirming what the technology is

424
00:27:38,240 --> 00:27:41,960
Speaker 1: doing for you. It's deciding what to do your business,

425
00:27:42,000 --> 00:27:45,600
Speaker 1: it's deciding the next holiday, it's deciding what you're going

426
00:27:45,680 --> 00:27:49,000
Speaker 1: to eat tomorrow or what type of workout you need

427
00:27:49,040 --> 00:27:52,120
Speaker 1: to use the technology as a tool to actually inform

428
00:27:52,640 --> 00:27:56,000
Speaker 1: you your decision other than the other way around.

429
00:27:56,080 --> 00:27:59,159
Speaker 2: Yeah, yes, sure, a lot to think about. That's in

430
00:27:59,240 --> 00:28:02,600
Speaker 2: a sailor a great pleasure having you thank you so

431
00:28:02,680 --> 00:28:05,840
Speaker 2: much for your time. That's chief Data and Analytics Officer

432
00:28:06,040 --> 00:28:09,440
Speaker 2: a different bee. Risk is also an author, a trailblazer

433
00:28:09,520 --> 00:28:12,399
Speaker 2: in the space doctor Mark Nacila, joining us this morning