Latest / Views and News with Clarence Ford / Can we slow down AI progress?
Transcript
- 0:01Speaker 1: I like that kind of competitive rivalry. Dean Makurubri joins us.
- 0:05Speaker 1: He's the chief AI strategy officer at Humana. We're going
- 0:10Speaker 1: to be asking the question. Can we slow down AI progress?
- 0:14Speaker 1: That's been demanded of late. We're talking AI leadership. We're
- 0:19Speaker 1: talking Dario Amodai, Sam Altman demanding a slowdown. There's Jakub Paczoki. Now,
- 0:28Speaker 1: these are the bosses. Of course, there's Ramaphosa's recent warning
- 0:32Speaker 1: at the BRICS summit. All of these coming together on
- 0:36Speaker 1: the same weekend. You just heard it being discussed yesterday.
- 0:39Speaker 1: in our regular BBC feature. Every single media is discussing
- 0:44Speaker 1: it around the world because some of this leadership are
- 0:47Speaker 1: suggesting that we, AI can take us out, can snuff
- 0:50Speaker 1: us out by the end of the decade. We got
- 0:54Speaker 1: three years left. That's the news that's come out last week.
- 0:58Speaker 1: And it's now a primary foreign policy and a corporate
- 1:02Speaker 1: governance issue, AI capabilities could become uncontrollable. Okay, so the dilemma,
- 1:08Speaker 1: I think, is the creators of technology are warning that
- 1:13Speaker 1: they might lose control of it. We can't be treating
- 1:16Speaker 1: the hesitation as a marketing ploy or economic concern. I
- 1:20Speaker 1: think that's going to be a very dangerous oversight. And
- 1:23Speaker 1: I guess at the same time, asking for a pause
- 1:25Speaker 1: in We're seeing that asymmetric, we were talking about asymmetric
- 1:28Speaker 1: global arms race. That's what's happening at this moment. We
- 1:30Speaker 1: spoke about it last week. It presents a massive, massive
- 1:35Speaker 1: game theory problem, I would propose as well. Welcome to you,
- 1:40Speaker 1: Dean McCoubry. We need to make sense of it, and
- 1:42Speaker 1: we're going to rely on you for that objective.
- 1:47Speaker 2: That's actually no small task, am I correct?
- 1:51Speaker 1: I would imagine. But the big guns are now saying,
- 1:54Speaker 1: hold on a second. Yeah, they are looking at themselves
- 1:59Speaker 1: and they're saying, hold on a second, I need to
- 2:01Speaker 1: pause here. It's not us saying, shouldn't you pause? Maybe
- 2:04Speaker 1: you should have another look at stuff. They're now saying it.
- 2:09Speaker 2: Yeah, let's go back a tiny bit. I'll tell you
- 2:11Speaker 2: where this all started. It was simmering in the last month.
- 2:17Speaker 2: So in the last five weeks, we've had a series
- 2:19Speaker 2: of announcements that have been a little bit startling. But
- 2:22Speaker 2: the bomb kind of went off on the 8th of
- 2:24Speaker 2: September when a guy called Jacob Coxon, who works at Anthropic,
- 2:29Speaker 2: basically decided to resign. And stated that we are racing
- 2:35Speaker 2: towards self-improving superintelligence, and we are gambling with our lives
- 2:40Speaker 2: by the end of the decade, which is about three
- 2:42Speaker 2: and a half years. So obviously, the world sat up
- 2:45Speaker 2: and listened, very similar to the way in which Tristan
- 2:48Speaker 2: Harris from the Center for Humane Technology raised a big
- 2:52Speaker 2: flag on social media with his movie, The Social Dilemma,
- 2:56Speaker 2: That was probably maybe six or seven years ago. It's
- 2:59Speaker 2: very similar to that. And then the world obviously sat
- 3:02Speaker 2: up and said, wow, this guy is actually leaving. And
- 3:06Speaker 2: I believe it was before his shares were to be vested.
- 3:08Speaker 2: It was about to make some money. So you can't
- 3:10Speaker 2: just say it was a publicity stunt. And so then now,
- 3:14Speaker 2: I think in the same day, out of 110 million
- 3:17Speaker 2: views on those comments, and then he started popping up
- 3:19Speaker 2: on CNN and Sky News and so on. So is
- 3:23Speaker 2: he right? Well, I think the reality is around AI,
- 3:26Speaker 2: and I've been saying this for years. I write for
- 3:28Speaker 2: Daily Maverick and so on. is that nobody really knows. Remember,
- 3:31Speaker 2: there's a lot of interventions that can happen between.
- 3:34Speaker 1: Now and then.
- 3:35Speaker 2: I mean, government could wake up. Regulators could wake up.
- 3:38Speaker 2: Big tech could decide to slow down, which is what
- 3:42Speaker 2: was suggested. So Bill Gates wrote a very long letter
- 3:45Speaker 2: all about kind of saying that we needed to really
- 3:49Speaker 2: pay attention to what was going on. That was on
- 3:53Speaker 2: the 26th of August. He wrote this long letter, people
- 3:55Speaker 2: should look it up, that we're in one of the
- 3:57Speaker 2: most turbulent periods in human history. And then just a
- 4:01Speaker 2: few days later, of course, Dario Amodeo also said, maybe
- 4:05Speaker 2: we should slow things down. Now, if we did slow
- 4:07Speaker 2: things down, and if governments did step in, there is,
- 4:11Speaker 2: of course, the potential that we could get a better
- 4:14Speaker 2: handle on Because at the moment, AI is advancing quicker
- 4:19Speaker 2: than the humans can handle it. So with those measures, possibly,
- 4:24Speaker 2: but do I believe those measures? Do I believe Gates
- 4:27Speaker 2: and Amodei, or is it just PR spin?
- 4:31Speaker 1: Yeah, and I think they're making very important calls, Dario, Amodei.
- 4:37Speaker 1: Jakub Paczoki, they essentially urging world leaders to treat AI oversight,
- 4:45Speaker 1: and I'm quoting, with the same mandatory rigor applied to aviation,
- 4:51Speaker 1: pharmaceuticals and nuclear power. That's like a no brainer, is
- 4:56Speaker 1: it not?
- 4:58Speaker 2: Yeah, it is a no-brainer, particularly because we've watched what's
- 5:01Speaker 2: happened with social media over the last 20 years, in
- 5:05Speaker 2: which it took them 20 years to get the first country, Australia,
- 5:10Speaker 2: to say, let's ban social media for anybody under 16.
- 5:16Speaker 2: 20 years, which is completely insane. And I've been teaching
- 5:19Speaker 2: kids around media for nine of those years and pleading
- 5:23Speaker 2: that governments get more involved in this. So.
- 5:27Speaker 2: Do I believe it? Not really so much because the
- 5:30Speaker 2: irony in the last four or five, and this is
- 5:33Speaker 2: the most important point here, is that in the last
- 5:36Speaker 2: four weeks, what also happened was that ChatGPT announced their
- 5:42Speaker 2: latest model called Astra. And also, I think they announced
- 5:47Speaker 2: Claude Fable 5.1, which are the most advanced models in
- 5:51Speaker 2: both those portfolios, both those LLM platforms, right? And soon
- 5:55Speaker 2: after that, we started to see a lot of concerns
- 5:58Speaker 2: around the fact that we might be seeing more cybersecurity
- 6:02Speaker 2: incidents because of the power of Astra. and 5.1. So
- 6:08Speaker 2: let me just go back. We've got these leaders in
- 6:10Speaker 2: big tech who clearly have the loudest voices and the
- 6:13Speaker 2: most money announcing Astra 6 and 5.1, but unable to
- 6:19Speaker 2: stop doing that, saying, we should put more regulation in,
- 6:22Speaker 2: you know. But there's not really like a community between
- 6:26Speaker 2: them where they could possibly agree because to some degree,
- 6:30Speaker 2: it's a race to the summit.
- 6:34Speaker 1: Yeah, I get you. And I think... A lot of people,
- 6:37Speaker 1: people like me, and I would suggest we by far
- 6:40Speaker 1: in the majority, we like just how realistic is a
- 6:45Speaker 1: kill switch? Why can't we just factor a kill switch
- 6:48Speaker 1: into the occasion? Then they can go on and multiply
- 6:52Speaker 1: and we can kill them whenever we need to.
- 6:55Speaker 2: A hundred percent, right? That would make the most sense.
- 6:58Speaker 2: But then what we need to do is we need
- 7:00Speaker 2: to think broader than just what's happening inside big tech.
- 7:04Speaker 2: We need to think about, for example, geopolitics. So we
- 7:07Speaker 2: need to think about the US versus China. Do you
- 7:09Speaker 2: think they're going to pull out of any type of
- 7:11Speaker 2: AI race right now? Because it also, of course, impacts
- 7:15Speaker 2: intelligence and cyber warfare. and all of those types of things.
- 7:19Speaker 2: So we've got to remember that there is almost this momentum,
- 7:22Speaker 2: this push, where everybody is competing to stay ahead because
- 7:26Speaker 2: what happens if you get behind? What does that do
- 7:29Speaker 2: to your economy? What does that do to your defense system?
- 7:32Speaker 2: What does that do to the potential for your utilities
- 7:35Speaker 2: to be broken down? Oh, no, we just need to
- 7:37Speaker 2: push on. So there's a difference between where does the
- 7:42Speaker 2: tail switch go, who owns it, So in which areas
- 7:46Speaker 2: should we put it, for example? You know, AI is
- 7:50Speaker 2: so broad. We've got generative AI, but we've got many
- 7:54Speaker 2: other forms of AI. The generative AI is the type
- 7:57Speaker 2: of stuff that we are using, for example. What about
- 8:00Speaker 2: the other forms of AI? Does that get included? And
- 8:03Speaker 2: who would make that decision? So do you think that
- 8:06Speaker 2: governments are going to stop because everybody's a little bit
- 8:10Speaker 2: worried around what Dario and Sam and Elon have to say?
- 8:15Speaker 2: I don't think so. Only they, truly, only they would
- 8:19Speaker 2: be able to get together and say, are we serving humanity?
- 8:24Speaker 2: Will we be the great villains of of this decade.
- 8:28Speaker 2: And people will look back and go, you really changed
- 8:32Speaker 2: the course of the world. And I think it would
- 8:34Speaker 2: be the onus and the responsibility is on them. Even
- 8:37Speaker 2: Bill Gates says this. He says, I would wholly support
- 8:41Speaker 2: a move to be able to regulate and contain AI
- 8:44Speaker 2: because it's moving faster than we can handle. He doesn't
- 8:49Speaker 2: say he's going to do it the way he did
- 8:51Speaker 2: with his philanthropic efforts and the work with TV.
- 8:56Speaker 1: Going to do it.
- 8:56Speaker 2: It's going to take someone to do it, and I
- 8:59Speaker 2: don't think government's going to be the one.
- 9:01Speaker 1: Okay, just the point that you made now, and I
- 9:03Speaker 1: think we need to understand the nature of the existential threat.
- 9:07Speaker 1: I think the question is, is it a rogue machine
- 9:12Speaker 1: wiping us out, or is it humanity willingly surrendering that
- 9:18Speaker 1: critical thinking, that agency, until we're just passive spectators?
- 9:25Speaker 2: I think it's so many things. Let's go to where
- 9:28Speaker 2: the mind goes to. So the mind, naturally, when we
- 9:31Speaker 2: hear these stories, we hear a headline. The media has
- 9:34Speaker 2: been doing this for 50, 60, 70 years. We hear
- 9:38Speaker 2: a headline, and that headline can sometimes, in many people,
- 9:41Speaker 2: not in everybody, go critical. catastrophizing, which goes to the end.
- 9:47Speaker 2: And suddenly we see those Chinese robots that can run
- 9:50Speaker 2: at lightning speed quicker than Usain Bolt. But they're carrying
- 9:52Speaker 2: a machine gun and they're cleaning out the high streets
- 9:55Speaker 2: in New York and the rest of the world. I'm
- 9:57Speaker 2: not being funny here. You know, we've all seen the
- 9:59Speaker 2: Terminator movies and so on. So what happens is we
- 10:02Speaker 2: might think that it's that. And you might be saying,
- 10:05Speaker 2: but hold on, you know, we're actually going to be
- 10:07Speaker 2: getting digital dementia here. because ultimately we've removed the agency
- 10:12Speaker 2: and the independent thought that we have. And now AI
- 10:15Speaker 2: just does stuff for us and so we're going to
- 10:17Speaker 2: be stupid. I'm not sure if it's that either. I
- 10:19Speaker 2: think the existential threat actually comes from areas like, you know,
- 10:26Speaker 2: things like cyber warfare and the things that are going
- 10:29Speaker 2: to happen between governments. And what happens if potentially AI...
- 10:38Speaker 2: When it takes on the role of advanced AI or AGI,
- 10:42Speaker 2: that it starts to move off in different directions without
- 10:45Speaker 2: us knowing. Now, that's just started to happen, which is
- 10:48Speaker 2: what happened with the hugging face scenario that's been in
- 10:51Speaker 2: the media recently. Unless it's constrained and it knows when
- 10:55Speaker 2: to stop and it knows what not to do and
- 10:59Speaker 2: it's very carefully defended, guarded, there are warnings and alerts
- 11:04Speaker 2: around that, then ultimately it could go off on its
- 11:07Speaker 2: own and do crazy stuff. And that could be bringing
- 11:10Speaker 2: down utilities, that could be crashing the banking system. You know,
- 11:14Speaker 2: that's when you start to go, hold on, what happens
- 11:16Speaker 2: to electricity? What happens to water? What happens to safety, etc.?
- 11:21Speaker 2: And I think that can create panic, which can then
- 11:24Speaker 2: create a lot of other problems.
- 11:25Speaker 1: And all of that can be incapacitated. by AI and
- 11:31Speaker 1: by an enemy. Let's understand that. Let's understand that because
- 11:37Speaker 1: we've got a lot of geopolitical tension happening at this
- 11:39Speaker 1: moment in time. And can a pause actually work? You know,
- 11:43Speaker 1: when you've got really significant national security races that would
- 11:50Speaker 1: kind of make slowing down or turn slowing down into
- 11:54Speaker 1: geopolitical suicide.
- 11:58Speaker 2: I mean, let's take a look at how we're doing
- 12:00Speaker 2: at the moment geopolitically. When we look at, let's just
- 12:04Speaker 2: look at, let's say, for example, the US, China, Russia,
- 12:09Speaker 2: the Middle East, what is happening at the moment? You know,
- 12:14Speaker 2: it's a case of definitely never the twain shall meet.
- 12:17Speaker 2: You know, it's really difficult for people to agree because
- 12:20Speaker 2: to some degree, there is already a physical land grab
- 12:24Speaker 2: for territory. This is really important. The world has always
- 12:28Speaker 2: been driven by physical land grabs for territory. It's power, dominance, greed,
- 12:35Speaker 2: the champion of the world type of thing. It's just
- 12:38Speaker 2: that now it's more virtual and it's more insidious, it's
- 12:41Speaker 2: more secretive. So ultimately, the question is, do I think
- 12:46Speaker 2: that China, Russia, the Middle East would necessarily play more?
- 12:51Speaker 2: I'm not sure because they don't manage to agree on
- 12:54Speaker 2: much at the moment. You know, it's just the same
- 12:57Speaker 2: way that you might say, do I think that Sam
- 13:00Speaker 2: Altman would listen to Dario Amodei and take a two-year
- 13:03Speaker 2: hiatus on the development of AI? I'm not sure. Are
- 13:07Speaker 2: we being fed spin by one party? Are some parties
- 13:11Speaker 2: conscious and other parties just pure greed and dominance? I
- 13:15Speaker 2: don't know. But if you look at data, if you
- 13:18Speaker 2: just look at data geopolitically, I don't think it's possible,
- 13:22Speaker 2: a global pause. Yes. Yes.
- 13:25Speaker 1: And that's the sad thing. So it's the human element
- 13:28Speaker 1: in the equation that is the big problem. Then, okay,
- 13:32Speaker 1: now let's talk about the frontier models. We're talking about
- 13:34Speaker 1: the latest innovation, the latest AI models that are out
- 13:38Speaker 1: there already. I'm wondering if a kill switch, because they've
- 13:43Speaker 1: got autonomous goal setting as part of their functions. A
- 13:48Speaker 1: kill switch... Doesn't apply to them, does it? Yeah.
- 13:57Speaker 2: Well, this is probably my other big thing that I...
- 14:04Speaker 2: want people to think about is we keep thinking that
- 14:08Speaker 2: AI is this thing, right? It's like AI. We already
- 14:12Speaker 2: see it to some degree as this rogue robot, but
- 14:15Speaker 2: it was born from humans. It's perpetuated by humans. The
- 14:21Speaker 2: shadow side of what it's capable of is driven by humans,
- 14:25Speaker 2: be it leaders, platform builders cyber criminals etc this is
- 14:31Speaker 2: coming from humans we have a human job to do
- 14:34Speaker 2: and regulators have a constraint job to do okay so
- 14:39Speaker 2: so do i think that there's a kill switch it
- 14:41Speaker 2: would need to come to conscience it would need to
- 14:44Speaker 2: come to leadership it would need to come to uh
- 14:47Speaker 2: the trumps and uh you know the the leader of
- 14:50Speaker 2: china i don't say his name very well she's jin
- 14:52Speaker 2: pin i think i She says that really badly. But
- 14:58Speaker 2: it comes to consciousness and leadership. It comes to foresight.
- 15:02Speaker 2: It needs to go beyond power. So Again, if you
- 15:07Speaker 2: look at humanity going back a thousand years, how have
- 15:09Speaker 2: we been doing with that as well? So I'm not
- 15:12Speaker 2: being negative that it's not possible, but the steps that
- 15:15Speaker 2: we'd need to take is that people would say, this
- 15:19Speaker 2: is advancing quicker than humanity can carry it. We're only
- 15:23Speaker 2: on AstraZeneca. GPT and Fable 5.1, and there's many others.
- 15:28Speaker 2: So what does this look like in 2028, 2029, 2030,
- 15:32Speaker 2: when we're looking at the types of threats that we're
- 15:36Speaker 2: seeing already in just three years of generative AI? And
- 15:41Speaker 2: the question about kill switches, who puts it in, when
- 15:45Speaker 2: does it get deployed, and so on? This is judgment
- 15:48Speaker 2: and leadership and conscience. Super tricky, right?
- 15:53Speaker 1: At a time when psychopaths and narcissists are in ascension.
- 15:56Speaker 1: That's scary, right?
- 15:58Speaker 2: Mm-hmm. It is. And I think that that's the thing
- 16:01Speaker 2: for me. I think about it and I think that
- 16:04Speaker 2: the idea of life skills and leadership and digital life
- 16:08Speaker 2: skills is so important now in society. The training, this
- 16:13Speaker 2: is what we do at Humane. We're an AI marketing
- 16:17Speaker 2: agency and we try and speak about, with very large enterprises, policy, ethics, guardrails, contexts.
- 16:26Speaker 2: And we help them take that entire journey to still
- 16:29Speaker 2: extract the efficiencies that AI has, do it with a
- 16:34Speaker 2: conscience and do it with leadership, deeply important. And we
- 16:37Speaker 2: need to do that in our schools as well. But unfortunately,
- 16:39Speaker 2: as we know, education is an oil tanker that's incredibly
- 16:43Speaker 2: slow to turn. And it's work that I'm just starting
- 16:46Speaker 2: to do more now. I'm speaking to the WCED in
- 16:50Speaker 2: about three weeks now. to speak to 250 teachers there
- 16:55Speaker 2: about their responsibility with the reality of AI. And then
- 17:00Speaker 2: a week after that, I'm going to Amsterdam for the
- 17:02Speaker 2: World AI Summit to speak with some of the biggest
- 17:04Speaker 2: names in the world on this exact topic that we're
- 17:07Speaker 2: talking about today. So thanks for preparing me, Clarence. You
- 17:11Speaker 2: helped me.
- 17:11Speaker 1: Yeah. We got to take the world back from the psychopaths,
- 17:16Speaker 1: from the narcissists. We got to get the The responsibility back,
- 17:20Speaker 1: that's a word that you've used. We've got to get
- 17:23Speaker 1: the morality back. We've got to get the ethics back.
- 17:30Speaker 1: This is the good fight. Because if we have people
- 17:35Speaker 1: of those kind of values at its helm, AI could
- 17:39Speaker 1: be so beneficial to the world and its future. Dean,
- 17:42Speaker 1: can we end on that point? We can.
- 17:45Speaker 2: I would say that leadership... from government into company, leadership
- 17:50Speaker 2: and family, or underpinned by the values of those families,
- 17:55Speaker 2: of those organizations, and of government, from top to bottom.
- 17:59Speaker 2: That's how we do it, but we have to try
- 18:01Speaker 2: and go back and fix that as well. If we
- 18:03Speaker 2: invest in that now, in five years, we might see
- 18:06Speaker 2: slightly different outcomes.
- 18:09Speaker 1: It's a good fight. Thank you, Dean. McCoubry, Chief AI
- 18:12Speaker 1: Strategy Officer at UMaine. And this is not an argument, again.
- 18:17Speaker 1: This is not about shouting louder. This is about raising
- 18:19Speaker 1: the vibration. This is about raising the vibration of those
- 18:23Speaker 1: bad scenarios that have been sketched in the past. We
- 18:25Speaker 1: could be a reality in the not-too-distant future. It is
- 18:30Speaker 1: two minutes from 11 o'clock on that very scary kind
- 18:34Speaker 1: of moment. And, of course, Natalie Malka standing by with
- 18:37Speaker 1: Eyewitness News. at the top of the hour. In the
- 18:41Speaker 1: last 24 hours, what's happened in your world? In a
- 18:45Speaker 1: Senate presidency, Senator Sue H. Cunga has appealed for urgent
- 18:48Speaker 1: action to protect women using public spaces following the discovery
- 18:52Speaker 1: of four bodies in Kempton Park. Police say the victims
- 18:54Speaker 1: were all found partially dressed and bruised in the area.
- 18:57Speaker 1: Officers have since confirmed a