Latest / INSIGHT TODAY / Musk, Zuckerberg and Altman clash over AI's future

Transcript

Musk, Zuckerberg and Altman clash over AI's future · INSIGHT TODAY

• 0:00
  Speaker 1: OpenAI chief Sam Altman is visiting Washington to meade White
• 0:03
  Speaker 1: House officials and lawmakers to discuss artificial intelligence policy. His
• 0:07
  Speaker 1: visit comes at a particularly sensitive moment, days after an
• 0:11
  Speaker 1: Open AI test model escaped a lab environment and hacked
• 0:14
  Speaker 1: into a different company, and it comes as staffers from
• 0:17
  Speaker 1: frontier AI companies call for a slowdown in AI development
• 0:22
  Speaker 1: in an open letter, well as intense debate rages over
• 0:25
  Speaker 1: how the US should regulate to the fast developing technology
• 0:28
  Speaker 1: and respond to Chinese competition. Mark Zuckerberg says the US
• 0:33
  Speaker 1: government should not block Chinese AI. The CEO of Meta
• 0:36
  Speaker 1: is also making the case for bringing superintelligence to more people.
• 0:41
  Speaker 1: In an article for The Wall Street Journal. He says
• 0:43
  Speaker 1: that that can boost innovation and empower people. CNN Tech
• 0:47
  Speaker 1: reporter Claire Duffy joins US Now from New York. So
• 0:50
  Speaker 1: much to break down here The timing of this op
• 0:53
  Speaker 1: ed from Mark Zuckerberg sam Altman's visit, each signaling something
• 0:59
  Speaker 1: quite different. Sam Altman quite concerned, perhaps just about what
• 1:03
  Speaker 1: happened with their model last week, and Mark Zuckerberg seems
• 1:09
  Speaker 1: to be envisioning a completely different future for AI. Just
• 1:14
  Speaker 1: talk to us about the timing and what both of
• 1:17
  Speaker 1: these men are arguing. Yeah, Beanna.
• 1:19
  Speaker 2: It's really interesting because Mark Zuckerberg seems to be trying
• 1:22
  Speaker 2: to shift the narrative around AI away from some of
• 1:25
  Speaker 2: these sort of doomerism scenarios that leaders in the industry
• 1:29
  Speaker 2: like Sam Altman have laid out here, and in large part,
• 1:32
  Speaker 2: I think Meta is trying to speak to the public
• 1:34
  Speaker 2: here and paint a rosier picture of what the AI
• 1:37
  Speaker 2: future could look like. In this op ed, Mark Zuckerberg wrote, quote,
• 1:41
  Speaker 2: humanity has witnessed many transformative advances. Each time there is
• 1:45
  Speaker 2: fear that people will be left behind, but each time
• 1:48
  Speaker 2: humanity has come out with more people sharing greater prosperity, health,
• 1:52
  Speaker 2: and freedom. We believe this will be true with AI
• 1:55
  Speaker 2: as well, and the abundance of the future can be
• 1:57
  Speaker 2: shared by everyone. Mark Zuckerberg is also really trying to
• 2:01
  Speaker 2: stake out Meta's position in this ongoing debate in the
• 2:04
  Speaker 2: industry about how accessible these advanced AI models should be
• 2:08
  Speaker 2: and really sort of not calling them out by name,
• 2:11
  Speaker 2: but directing attention to leaders in this industry like Open
• 2:14
  Speaker 2: Ai and Anthropic that have advocated for restricting access to
• 2:18
  Speaker 2: the most advanced AI models. That is part of what
• 2:21
  Speaker 2: Altman and others are in Washington this week to talk about.
• 2:24
  Speaker 2: But Mark Zuckerberg said that he believes that's the wrong
• 2:28
  Speaker 2: path and that we should be focused on open access
• 2:30
  Speaker 2: to this technology. I'll just read you another portion of
• 2:33
  Speaker 2: his piece. He says, if only a handful of institutions
• 2:37
  Speaker 2: have super intelligence, they will inevitably exercise a controlling influence
• 2:41
  Speaker 2: over economics, science, and politics, even with the best intentions.
• 2:45
  Speaker 2: That concentration would limit people's ability to choose their own future.
• 2:49
  Speaker 2: But it's interesting, Biana, because on one hand, Mark Zuckerberg
• 2:52
  Speaker 2: is saying there needs to be open access to this technology.
• 2:55
  Speaker 2: On the other hand, Meta this year just released its
• 2:58
  Speaker 2: first closed source AI Monde, which restricts what people can
• 3:01
  Speaker 2: do with it, after Zuckerberg had previously advocated for open
• 3:05
  Speaker 2: source technology. So there is some tension, I think, between
• 3:08
  Speaker 2: what he's saying and the direction that the company has
• 3:10
  Speaker 2: been going in. But I think the other important part,
• 3:13
  Speaker 2: and the other important piece of the timing here for
• 3:15
  Speaker 2: Meta in particular, is the company is reporting earnings tonight.
• 3:19
  Speaker 2: The shares of the company are down about eight percent
• 3:21
  Speaker 2: since the start of this year because investors are asking
• 3:24
  Speaker 2: tough questions about how much money Meta is spending on
• 3:27
  Speaker 2: AI infrastructure. They want to see a greater return from
• 3:30
  Speaker 2: the company, So I think in general it makes sense
• 3:32
  Speaker 2: that Meta is in this moment trying to paint a
• 3:35
  Speaker 2: rosier picture around AI than perhaps some of his counterparts are,
• 3:39
  Speaker 2: who are really concerned about the risks.
• 3:41
  Speaker 1: And it's worth asking, given the timing here, and given
• 3:44
  Speaker 1: the narrative that Mark Zuckerberg is now trying to promote,
• 3:48
  Speaker 1: perhaps they're just hoping to buy time as well for
• 3:52
  Speaker 1: investors and with investors, as their open models appear to
• 3:55
  Speaker 1: be lagging the frontier. At this moment, all things can
• 3:58
  Speaker 1: be true. I guess Duffy, thank you well. One more
• 4:03
  Speaker 1: voice is weighing in over this debate on the future
• 4:05
  Speaker 1: of AI. You know this man Elon Musk, was asked
• 4:08
  Speaker 1: whether human beings would be in control in ten years time. Musk,
• 4:12
  Speaker 1: of course, is one of the worlds. He is the
• 4:14
  Speaker 1: world's richest man and the CEO of Tesla and SpaceX,
• 4:17
  Speaker 1: his rocket and AI company. Take a listen to how
• 4:20
  Speaker 1: he answered a question from the editor in chief of
• 4:23
  Speaker 1: The Economist.
• 4:25
  Speaker 3: You're sure that humans will no longer be in control
• 4:28
  Speaker 3: in tenya's time?
• 4:29
  Speaker 4: I think it is unlikely in the same way that
• 4:33
  Speaker 4: if the difference in intelligence between AI and humans is
• 4:38
  Speaker 4: fast and greater than the difference in intelligence between AI
• 4:40
  Speaker 4: and chimpanzees, it's hard to imagine that the chimpanzees would
• 4:45
  Speaker 4: be in charge. All right.
• 4:47
  Speaker 1: On that note, time for the exchange where we want
• 4:49
  Speaker 1: to dig deeper into the world of artificial intelligence. Joining
• 4:52
  Speaker 1: me now Sebastian Mallaby, journalist and Senior Fellow for International
• 4:55
  Speaker 1: Economics and the Council and Form Relations. We should note
• 4:59
  Speaker 1: that that interview with Elon Musk was not only conducted
• 5:02
  Speaker 1: with the Economist editor in chief, the brilliant Zandy Minobettos,
• 5:05
  Speaker 1: but she's also your wife, so this seems to be
• 5:08
  Speaker 1: a family affair. It is a fascinating interview. It is
• 5:12
  Speaker 1: worth the time to listen to all of his points
• 5:16
  Speaker 1: and thoughts on a number of issues as it relates
• 5:19
  Speaker 1: to AI. I was fascinated to see how he's really
• 5:23
  Speaker 1: made an arc in being very concerned about it being
• 5:27
  Speaker 1: an existential threat to perhaps now coming and being at
• 5:30
  Speaker 1: peace with the fact that, as he said, at least
• 5:33
  Speaker 1: in his version of a fact, that in ten years time,
• 5:36
  Speaker 1: humans will not be in charge. He says while monkeys
• 5:40
  Speaker 1: seem to be quite happy swinging from trees. I don't
• 5:43
  Speaker 1: know if that's the analogy that most people will be
• 5:45
  Speaker 1: at peace with. But here you have Zuckerberg making the
• 5:48
  Speaker 1: mirror image point in his op ed, and he's saying
• 5:51
  Speaker 1: he doesn't understand why anyone who believes AI would perhaps
• 5:55
  Speaker 1: be an existential threat or eliminate humans relevance would rush
• 5:59
  Speaker 1: to build to that future. So how do you square
• 6:01
  Speaker 1: those two arguments?
• 6:04
  Speaker 3: You know, I think they've both got much to be
• 6:07
  Speaker 3: you know, many gaps in them. Right, So but Mark
• 6:10
  Speaker 3: Zuckerberg view that this is going to be great. You know,
• 6:13
  Speaker 3: on balance, new technologies have been great, but they also
• 6:16
  Speaker 3: have downsides. AI has very clear risks attached to it.
• 6:20
  Speaker 3: So the sensible position is to be both optimistic and pessimistic. Right,
• 6:23
  Speaker 3: it can do great things. It can advance science, it
• 6:25
  Speaker 3: can empower people to find out new stuff. But it
• 6:28
  Speaker 3: can also be used by bad actors for cyber hacks,
• 6:30
  Speaker 3: for bioterrorism in the future. Perhaps there's even an autonomy risks.
• 6:34
  Speaker 3: There's deep fake with this, all these risks. So to
• 6:37
  Speaker 3: sit there happily and say, don't worry, guys, it's all
• 6:39
  Speaker 3: going to be good. I mean, it's completely not convincing.
• 6:42
  Speaker 3: On the other hand, the Elon Musk view, which is
• 6:44
  Speaker 3: we can't control it, but hey, so what, that's just
• 6:48
  Speaker 3: not too reassuring either. And in fact, when you know
• 6:52
  Speaker 3: at the dawn of the AI race, you know, back
• 6:54
  Speaker 3: in around twenty fourteen, there was this famous argument between
• 6:58
  Speaker 3: Elon Musk and Page of Google where Larry Page is like,
• 7:03
  Speaker 3: it's fine, AI will take over. You know, humans will
• 7:05
  Speaker 3: be eclipsed as long as intelligence itself is preserved. It
• 7:09
  Speaker 3: can be preserved on silicon who cares about biological intelligence
• 7:12
  Speaker 3: aka humans? And Elon flips at this, and you know,
• 7:17
  Speaker 3: they have a massive bust up. They had been close
• 7:19
  Speaker 3: friends and they became sort of enemies over this debate,
• 7:22
  Speaker 3: and now Edon is saying, hey, cool, don't worry about it.
• 7:24
  Speaker 3: I mean, it is a weird shift.
• 7:27
  Speaker 1: It is indeed, And going back to Zuckerberg's point in
• 7:32
  Speaker 1: the conclusion he's come to, he's saying the concentration actually
• 7:35
  Speaker 1: of this power is the real threat. And he wrote
• 7:38
  Speaker 1: in his op ed the balance of power should be
• 7:40
  Speaker 1: the foundation of safety. And one of the thought experiments
• 7:44
  Speaker 1: he uses as a person with a super intelligent lawyer
• 7:47
  Speaker 1: is dangerous, but somebody in court also with a super
• 7:51
  Speaker 1: intelligent lawyer would be more equal and just. And he
• 7:55
  Speaker 1: worked about regulatory capture. If labs like open AI and
• 7:59
  Speaker 1: and threat Topic gets peer get to peer review the
• 8:03
  Speaker 1: industry and no one else, do you think that's a
• 8:05
  Speaker 1: principal case for democratizing this technology.
• 8:11
  Speaker 3: No, no, I don't. I think it's rather ironic when
• 8:13
  Speaker 3: you know, somebody goes to Washington, is photographed with the president,
• 8:18
  Speaker 3: you know, and publishes an op ed in the Wall
• 8:20
  Speaker 3: Street Journal and gives an ind to the Financial Times,
• 8:23
  Speaker 3: you know, and then says, somebody else is doing regulatory capture.
• 8:26
  Speaker 3: I mean, you know who's doing who's capturing? What I mean?
• 8:30
  Speaker 3: I mean, you know, clearly Meta has a big Washington
• 8:33
  Speaker 3: lobbying operation, just like the other labs do, and they're
• 8:36
  Speaker 3: all trying to influence policy, and that's how the game
• 8:38
  Speaker 3: is played. But I think, you know, pointing at the
• 8:40
  Speaker 3: other guys and saying they're doing capture and I'm not
• 8:43
  Speaker 3: is a bit is a bit ridiculous. So yeah, this comes.
• 8:48
  Speaker 1: As we noted, Sam Altman is also in Washington, perhaps
• 8:51
  Speaker 1: will be even meeting with the president, and uh, the
• 8:56
  Speaker 1: Open AI the safety test that they ran, which then
• 8:59
  Speaker 1: ultimately released this model, which they were not intending to.
• 9:04
  Speaker 1: It was a sealed off environment, it had no access
• 9:06
  Speaker 1: to internet, It found its way out of the sandbox,
• 9:09
  Speaker 1: got online, hacked into another tech company, and still answers
• 9:12
  Speaker 1: to a test that it was being given OpenAI said
• 9:14
  Speaker 1: that this was just hyper focused on passing this test,
• 9:18
  Speaker 1: that this perhaps is in indicative of a larger scale problem.
• 9:22
  Speaker 1: But what does that tell us about what this kind
• 9:24
  Speaker 1: of technology can do? Because something else I was struck
• 9:26
  Speaker 1: by in this interview with Zanny and Elon Musk was
• 9:30
  Speaker 1: in his timeframe, even five or six months from now
• 9:34
  Speaker 1: is a long time. I mean, most executives don't think
• 9:37
  Speaker 1: in five or six month increments. They're thinking five ten years.
• 9:40
  Speaker 1: The fact that we may see massive innovation in just
• 9:44
  Speaker 1: five months tells us what.
• 9:47
  Speaker 3: Yeah, Look, I mean, this technology is progressing extremely quickly.
• 9:51
  Speaker 3: And you know, one can't stress that enough because I
• 9:53
  Speaker 3: think people sometimes think, oh, you know, CHATTYBT arrived in
• 9:57
  Speaker 3: twenty twenty two, and then we had AI and it
• 9:59
  Speaker 3: was kind of one and done. No no, no, no,
• 10:01
  Speaker 3: it's not one and done. This thing changes every six
• 10:03
  Speaker 3: months and you get new capabilities. You get the ability
• 10:06
  Speaker 3: for these systems to have very long memories, to deal
• 10:09
  Speaker 3: with audio and pictures, to reason, to contribute frontier contributions
• 10:13
  Speaker 3: to mathematics, to act as agents. All these different things
• 10:17
  Speaker 3: have come out just in the space of you know,
• 10:19
  Speaker 3: less than four years, and the next couple of years
• 10:22
  Speaker 3: are going to be even wilder. If you listen to
• 10:24
  Speaker 3: dari amadavanthropic and I've got no reason to believe, you know,
• 10:28
  Speaker 3: he's saying anything other than the truth. And I've had,
• 10:30
  Speaker 3: by the way, the same thing kind of privately from
• 10:32
  Speaker 3: other lab leaders that you know, in about a couple
• 10:36
  Speaker 3: of years, you'll be able to tell a frontier AI system, hey,
• 10:40
  Speaker 3: go make the next version of yourself, which will be
• 10:43
  Speaker 3: much stronger, and the existing system will go off autonomously
• 10:47
  Speaker 3: and create the new system. This is called recursive self improvement,
• 10:50
  Speaker 3: and we're kind of two years away from that. It's
• 10:53
  Speaker 3: very serious.