1
00:00:02,960 --> 00:00:07,197
So, just to back up a bit, I've been
working on AI for for 10 years and

2
00:00:07,280 --> 00:00:10,997
probably the thing I've noticed most
about it is is how fast it's making

3
00:00:11,080 --> 00:00:14,637
progress. 2 years ago, it was at the
level of a smart high school student.

4
00:00:14,720 --> 00:00:17,877
Now, it's probably the level of a smart
college student and reaching beyond

5
00:00:17,960 --> 00:00:22,277
that. And on one hand, I think there's
a number of very positive things that are

6
00:00:22,360 --> 00:00:26,117
going to happen. I used to be a biologist.
I think AI has exactly the

7
00:00:26,200 --> 00:00:30,437
kind of skills that are needed to cure
important diseases like cancer and

8
00:00:30,520 --> 00:00:32,997
Alzheimer's, to provide cheaper energy,

9
00:00:33,080 --> 00:00:35,637
many positive things. But exactly those

10
00:00:35,720 --> 00:00:38,077
same kind of skills, things like

11
00:00:38,160 --> 00:00:40,797
summarizing a document, brainstorming,

12
00:00:40,880 --> 00:00:42,837
putting together a financial report,

13
00:00:42,920 --> 00:00:45,637
makes me worry a lot that entry-level

14
00:00:45,720 --> 00:00:49,797
jobs in areas like finance, consulting,

15
00:00:49,880 --> 00:00:53,077
tech, many many other areas like that,

16
00:00:53,160 --> 00:00:55,917
entry-level white-collar work, I worry

17
00:00:56,000 --> 00:01:00,557
that those things are going to be first
augmented but before long replaced by AI

18
00:01:00,640 --> 00:01:04,237
systems and that we we may
indeed it's hard to predict the

19
00:01:04,320 --> 00:01:08,397
future, but we may indeed have a have
a serious employment crisis on our hands

20
00:01:08,480 --> 00:01:10,797
as the pipeline for for these

21
00:01:10,880 --> 00:01:13,477
early-stage white-collar work starts to

22
00:01:13,560 --> 00:01:15,757
starts to contract and dry up. What's

23
00:01:15,840 --> 00:01:18,557
the timeline here?

24
00:01:18,640 --> 00:01:20,597
Yeah, so it's very hard to predict these

25
00:01:20,680 --> 00:01:25,677
things given how fast you know,
given how fast AI is making progress.

26
00:01:25,760 --> 00:01:30,597
But I would I would not be surprised
if somewhere between one and five years

27
00:01:30,680 --> 00:01:34,477
we started to see big effects here. You
know, I've heard a number of people talk

28
00:01:34,560 --> 00:01:36,677
about this in private, AI CEOs talk

29
00:01:36,760 --> 00:01:39,037
about this in private, CEOs of other

30
00:01:39,120 --> 00:01:42,077
companies talk about this in private.

31
00:01:42,160 --> 00:01:44,277
And I I really felt that you know, the

32
00:01:44,360 --> 00:01:49,117
message that this is happening hasn't
been getting out to to ordinary people,

33
00:01:49,200 --> 00:01:52,637
hasn't been getting out to our
legislators, our congress people either.

34
00:01:52,720 --> 00:01:56,877
And so I felt I needed to speak up
on the record. I do think that this is

35
00:01:56,960 --> 00:01:59,477
something we can prevent this,

36
00:01:59,560 --> 00:02:01,757
but we need to act now. I don't think we

37
00:02:01,840 --> 00:02:04,757
can stop the AI bus. You know, there are

38
00:02:04,840 --> 00:02:09,037
there are six or seven companies
just in the US working in this

39
00:02:09,120 --> 00:02:12,997
working in this area. I just run one
of them. Even if our company stopped doing

40
00:02:13,080 --> 00:02:16,717
what it was doing today, all the other
companies would would continue. Even if

41
00:02:16,800 --> 00:02:20,117
all six companies started, then China
would beat us. I think that's a big and

42
00:02:20,200 --> 00:02:24,157
important threat. So, we can't stop
the bus, but I do think we may have an

43
00:02:24,240 --> 00:02:26,717
opportunity to steer it. And and I think

44
00:02:26,800 --> 00:02:31,477
you know, we need to get congress
people, legislators, the public to to

45
00:02:31,560 --> 00:02:35,037
think about this now. One of the first
steps we're taking is just measuring the

46
00:02:35,120 --> 00:02:39,117
effects. Anthropic has put out something
that we call our economic index and

47
00:02:39,200 --> 00:02:43,717
using the economic index, we try and track
the rate at which these things are

48
00:02:43,800 --> 00:02:47,197
happening. You know, at some point
we have to go beyond measuring this to

49
00:02:47,280 --> 00:02:49,397
actually taking action, trying to build

50
00:02:49,480 --> 00:02:54,517
our AI systems in a way that augments
rather than replaces. But that won't be

51
00:02:54,600 --> 00:02:58,117
a long-term solution either. We need
to think about what what the government is

52
00:02:58,200 --> 00:03:02,917
going to do. And you know, there's a wide
range of actions that that the government

53
00:03:03,000 --> 00:03:06,437
the government could take. And you know,
it's it's not not my place to decide

54
00:03:06,520 --> 00:03:09,757
which of those are which of those are
correct. But but I think the first step

55
00:03:09,840 --> 00:03:13,157
is people need to think about this and
start taking it very seriously. Yeah,

56
00:03:13,240 --> 00:03:15,357
and you're talking about AI possibly

57
00:03:15,440 --> 00:03:18,077
wiping out all entry-level white-collar

58
00:03:18,160 --> 00:03:20,877
jobs and spiking unemployment to 10 to

59
00:03:20,960 --> 00:03:25,517
20%. So, dive into the mitigation
of this a little bit more. How could we

60
00:03:25,600 --> 00:03:29,717
mitigate the worst possible scenario here?

61
00:03:29,800 --> 00:03:32,837
Yeah, so it's hard to know the exact
numbers. I mean, you know, I don't have

62
00:03:32,920 --> 00:03:36,717
a crystal ball. I can't I can't predict
the future. But what I will say is I

63
00:03:36,800 --> 00:03:40,917
think what AI can do, you know, there
have been these technological changes in

64
00:03:41,000 --> 00:03:45,597
the past, but I think what's different
about AI is it's broader, it's deeper,

65
00:03:45,680 --> 00:03:49,757
and it's moving faster than than
previous such technologies. And so in

66
00:03:49,840 --> 00:03:54,317
terms of mitigations, you know, I think
I think one is just educating people in

67
00:03:54,400 --> 00:03:58,237
order in order to you in order to use AI.
There will always be people who are

68
00:03:58,320 --> 00:04:00,517
who are who are using AI and helping our

69
00:04:00,600 --> 00:04:03,317
workers to adapt faster. I think I think

70
00:04:03,400 --> 00:04:05,797
is an important move. A second move is

71
00:04:05,880 --> 00:04:10,597
that we may want we may want government
to you know, to to to kind of to kind of

72
00:04:10,680 --> 00:04:14,677
find a way to level the economic playing
field. And you know, this this may be a

73
00:04:14,760 --> 00:04:18,957
controversial proposal, but you know,
something like taxing AI companies like

74
00:04:19,040 --> 00:04:21,197
us. We we may need to get
to things that are actually

75
00:04:21,280 --> 00:04:25,157
>> message coming from you who runs
an AI company. But you're doing a public

76
00:04:25,240 --> 00:04:28,757
service announcement for all of us
in terms of this future concern. Dario,

77
00:04:28,840 --> 00:04:32,480
thank you so much for joining us. I'm sure