1
00:00:00,000 --> 00:00:01,757
Do you see a risk [music]

2
00:00:01,840 --> 00:00:04,797
in our seeding decision-making control

3
00:00:04,880 --> 00:00:09,197
to large, powerful AI models?
>> The The risk you've outlined is

4
00:00:09,280 --> 00:00:13,680
architecturally controllable.

5
00:00:15,120 --> 00:00:17,717
>> There is some authentic concern out

6
00:00:17,800 --> 00:00:20,397
there um which I wonder whether you I

7
00:00:20,480 --> 00:00:25,197
sympathize with, empathize with,
have any consideration for the risks of

8
00:00:25,280 --> 00:00:27,197
outsourcing data analysis and

9
00:00:27,280 --> 00:00:29,957
decision-making based on data analysis

10
00:00:30,040 --> 00:00:32,837
to a machine algorithm. Do you see a

11
00:00:32,920 --> 00:00:35,557
risk in our seeding decision-making

12
00:00:35,640 --> 00:00:40,157
control to large, powerful AI models who

13
00:00:40,240 --> 00:00:43,437
of course, while smart, are not human?

14
00:00:43,520 --> 00:00:47,877
>> Um the Look, the there are there's a
the general risk of what happens when the

15
00:00:47,960 --> 00:00:50,957
computer takes over. Then there's
the intermediate risk of which is what

16
00:00:51,040 --> 00:00:55,197
you're talking about. But But again,
the the risk you've outlined is

17
00:00:55,280 --> 00:00:58,237
architecturally controllable.

18
00:00:58,320 --> 00:01:02,477
Your Your enterprise, for example,
our product will allow you to granularly

19
00:01:02,560 --> 00:01:06,077
control what things are controlled by
the computer, what things are controlled

20
00:01:06,160 --> 00:01:08,917
by algorithms.
There's something in software called a

21
00:01:09,000 --> 00:01:12,837
write function, meaning who gets to have
the decision. This can be controlled

22
00:01:12,920 --> 00:01:16,677
down to this individual,
sub-individual in this role.

23
00:01:16,760 --> 00:01:18,237
But you need the right product to do it.

24
00:01:18,320 --> 00:01:21,117
And so a lot of the Not all, but a lot

25
00:01:21,200 --> 00:01:23,477
of the concerns in this area are because

26
00:01:23,560 --> 00:01:25,997
your software is software that was built

27
00:01:26,080 --> 00:01:28,557
for a world that was non-disjointed,

28
00:01:28,640 --> 00:01:33,277
peaceful, where your software product
was primarily sales, where you were not

29
00:01:33,360 --> 00:01:37,037
had did not have an enterprise where
software and software-defined products,

30
00:01:37,120 --> 00:01:39,357
both in military and commercial, powered

31
00:01:39,440 --> 00:01:42,317
to to some extent by AI, where that was

32
00:01:42,400 --> 00:01:46,757
not the case. And so one of the big
problems with reducing everything to a

33
00:01:46,840 --> 00:01:48,957
moral discussion is

34
00:01:49,040 --> 00:01:51,517
in many cases, the moral discussion is

35
00:01:51,600 --> 00:01:55,157
actually masks in intellectual and

36
00:01:55,240 --> 00:01:57,957
commercial engineering ineptitude. So,

37
00:01:58,040 --> 00:02:02,397
it's like these are These are
These are engineering problems.

38
00:02:02,480 --> 00:02:05,677
>> So, is this about messaging then?
Is this about communicating

39
00:02:05,760 --> 00:02:09,277
>> No, it's about the engineering actually
working. It's not a messaging problem.

40
00:02:09,360 --> 00:02:11,677
It's like And that's again why a small

41
00:02:11,760 --> 00:02:14,477
culture, whether it's Silicon Valley or

42
00:02:14,560 --> 00:02:17,037
the Emirates, massively outperforms a

43
00:02:17,120 --> 00:02:19,237
larger culture because

44
00:02:19,320 --> 00:02:22,597
there are moral and and equity issues

45
00:02:22,680 --> 00:02:25,077
here, no doubt. But a lot of the issues

46
00:02:25,160 --> 00:02:27,357
we pretend are moral are actually

47
00:02:27,440 --> 00:02:30,717
engineering. The Who controls the AI?

48
00:02:30,800 --> 00:02:33,717
Under what conditions
can it write to something?

49
00:02:33,800 --> 00:02:37,677
Uh Who What kind of decisions can be
made on the battlefield? Who gets to

50
00:02:37,760 --> 00:02:42,277
decide between life and death?
Is very similar, it not morally, not

51
00:02:42,360 --> 00:02:46,797
intellectually, but technically, who
controls where the satellite flies? Who

52
00:02:46,880 --> 00:02:50,437
tasks the satellite? Is it a human?
Is it a computer? Under what conditions

53
00:02:50,520 --> 00:02:53,397
should it be? Well,
there's a lot It depends the

54
00:02:53,480 --> 00:02:55,797
context. So, you know, if you're running

55
00:02:55,880 --> 00:03:00,437
a massive agricultural arm,
you probably want the

56
00:03:00,520 --> 00:03:04,477
business person to say, "I want
to expand this amount of resources on

57
00:03:04,560 --> 00:03:08,677
satellite coverage." In return, that
that's not a moral decision. It's a It's

58
00:03:08,760 --> 00:03:10,197
a competence decision. If you make it

59
00:03:10,280 --> 00:03:16,440
incorrectly, you will be defeated
by the person installing a better product.