Latest / CXOTalk / Big Tech's AI Oligopoly: Lords Reveal the Power Grab | CXOTalk 899
Transcript
- 0:00Microsoft has invested $13 billion in open AI.
- 0:05Amazon poured billions into Anthropic.
- 0:08Google controls your search data.
- 0:10Meta has your social graph. The AI oligopoly is here today
- 0:17on CXO talk number 899. Lord Tim Clement Jones and Lord
- 0:24Chris Holmes, key leader shaping UKAI policy in the House of
- 0:29Lords, reveal what business leaders must know.
- 0:32Now I'm your host, Michael Krigsman.
- 0:35Let's get into it. Tim.
- 0:37Tell us about your work in the House of Lords and your focus
- 0:41areas. I speak for my party on all
- 0:46matters sound, innovation and technology, and I chair a cross
- 0:51party group on artificial intelligence as well, which I
- 0:56founded back in 2016. Seventeen.
- 0:59And if one of the key issues for me is the power of big tech,
- 1:04which really is increasingly, it's like going back in in a
- 1:09sense to the big oil issues back in the 19th century where you
- 1:14had vertical integration of an absolutely essential utility by
- 1:20a limited number of businesses. And that's what we are seeing
- 1:24increasingly. So already we had a lot, a lot
- 1:27of big tech concentration. We've got, you know, control
- 1:30over 2 platforms really, Apple, Google.
- 1:34We've got cloud services really in the hands of say 2, Microsoft
- 1:38and Amazon Web Services. We've got concentration.
- 1:42We've got Facebook really, which is along with side TikTok, one
- 1:47of two really major social media outlets.
- 1:51So we're getting this concentration in digital, which
- 1:54is now being replicated in artificial intelligence and
- 1:58especially with the links between Open AI, for instance,
- 2:02and Microsoft, between Amazon and Anthropic, and of course
- 2:07Meta with its own large language model.
- 2:11So this kind of concentration is beginning to impact in a major
- 2:16way, which I think will have knock on effects for society,
- 2:20for our democracy, for countries trying to develop their own AI
- 2:25models. Chris, tell us about your work
- 2:29in the House of Lords and any initial thoughts that you have
- 2:32on this big tech set of issues. I try and cover new technologies
- 2:38including AI, blockchain, cyber and all things digital assets
- 2:43and tokenization with threads of ESG and EDI running through all
- 2:49of those. Also cover financial services
- 2:51regulation and all things fintech.
- 2:55And really what I'm trying to bring out are the possibilities,
- 3:00the potential of all of these new technologies of themselves.
- 3:05And then looking at some of the really exciting value cases
- 3:10where you look at the technologies in combination with
- 3:13one another. But we only win with these
- 3:17technologies if we human lead on them.
- 3:20We only win if we focus on the public good, the common good,
- 3:26and see them as enablers, empowers and really in all our
- 3:33human hands. That's the opportunity.
- 3:35But as we all know about opportunities, they're never
- 3:38inevitabilities is down to us to make it so.
- 3:41Tim just described this concentration of Microsoft and
- 3:47Open AI, Amazon, Anthropic, Google, Gemini and DeepMind.
- 3:52You have Meta with Mama. What is this concentration mean
- 3:58for competition and for society? It means concentration of
- 4:05information sources effectively. I mean, one of the the key
- 4:09things this isn't just about, you know, people accessing large
- 4:15language models, you know, for the hell of it and and for use
- 4:18in their daily life. And I entirely agree with Chris.
- 4:21The opportunities are there, but the risk is, is that what we're
- 4:25doing is we're getting our information from very limited
- 4:28sources. I mean, for instance, you know,
- 4:30the impact on news media, uh, of the digital era has been
- 4:36considerable with the monopoly of advertising that is now
- 4:40pretty much held by Google and Facebook.
- 4:43But of course, with the advent of artificial intelligence,
- 4:46which, you know, in a sense scrapes news data from the
- 4:50Internet, that means that concentration and that impact on
- 4:54the news media is even higher. So it's, this is becoming of
- 5:00great importance in terms of how people see the world and people
- 5:04grow up and they get all their information from one of the
- 5:08large language models, then they are directly influencing what
- 5:12we're doing and seeing. And I think that, you know, just
- 5:15digressing slightly, but robotics is going to pronounce
- 5:19that even more when AI is embedded in robotics that we use
- 5:25for everyday life. So this this is not a small
- 5:28thing. There's no way that we can say
- 5:31that we win with these technologies if we just have a
- 5:38winner takes all reality. There's no question these
- 5:43companies, we all know the list of them you set out at the
- 5:46outset, Michael. They've come an extraordinarily
- 5:50long way in a relatively short space of time.
- 5:55Their business model is clear, Their scaling has been clear,
- 6:00and that is 1 particular model, 1 particular approach.
- 6:06But we win with these technologies if we truly have a
- 6:11plurality and understanding the AI isn't 1 technology.
- 6:16AI isn't just LLM, which I think the public narrative is tending
- 6:22towards that sense. We've been on an AI journey and
- 6:25we've now reached this point and this is it.
- 6:27This is the Zenith in many ways, large language models, that's
- 6:30where we've been shooting towards.
- 6:32That's but one element of AI in a whole constellation of
- 6:36technologies. So I think each individual, each
- 6:40company, each community, each country needs to consider what
- 6:45do they want their AI? What do they want their
- 6:49technology story, their technology narrative be?
- 6:54How do they play to the elements, the values, the
- 6:58beliefs, the social, the democratic structures in that
- 7:01nation and really thinking deeply into this.
- 7:04And that will be the way that we succeed within this.
- 7:07It can't be that we just go down a route of winner takes all.
- 7:12Chris is absolutely right. Increasingly, I think, and this
- 7:16is if we're not careful, going to lead to a kind of nationalism
- 7:21as far as AI is concerned. There is this very strong mood
- 7:27both, you know, in countries within the European Union but
- 7:32also in the UK that we need to be able to develop our own
- 7:36sovereign AI capabilities and that if we're not careful, U.S.
- 7:40companies are going to drown out our own ability to develop AI
- 7:46tools of, you know, every description.
- 7:49And so this, if we're not careful, this is going to be a
- 7:52backlash. And, you know, I want to avoid
- 7:55that because I think we need to make sure that we take advantage
- 7:59of, you know, every kind of tool that is available
- 8:03internationally. But if we're dominated by three
- 8:07or four businesses, then I think that's going to create a
- 8:11reaction which would be extremely unhelpful.
- 8:14And you know, or we've seen what happens if you start applying
- 8:19sanctions and tariffs and so on. People get rather excited about
- 8:23that. And I don't believe that that is
- 8:25going to help develop our, if you like, technological
- 8:28partnerships across the world. Folks, we have two tremendous
- 8:34guests today. Ask questions if you are
- 8:39watching on LinkedIn, O your question into the LinkedIn
- 8:44comments, if you're watching on Twitter, use the #cxotalk.
- 8:49And we have some questions that are coming in already relating
- 8:52to the international aspects and the economic aspects.
- 8:56And we're going to get to those questions momentarily.
- 8:59But but I have a question for you both.
- 9:02You've both raised certain types of cautions.
- 9:06You've described the potential very significant impacts on
- 9:10society. But there's also another side to
- 9:13this, which is these companies are taking the risks they that
- 9:19you organizations have the capability to take in order to
- 9:24make these investments in the LLMS, in the data centres and
- 9:29infrastructure and so forth. And so frankly, given that
- 9:33they're making the investment, why, what business is it of any
- 9:39of us to interfere? Let them do it.
- 9:42Let us all reap the benefits of AI.
- 9:44History tells us that no industry sector, no market, no
- 9:55society or economy is at its best if there are no rules, if
- 10:02there are no regulations, if there isn't a sense of
- 10:06certainty, consistency, coherence, stability.
- 10:12Now with some way off that, I think with these current
- 10:18approaches to AI, but everything tells me from that history that
- 10:24right size regulation would be extraordinarily beneficial, not
- 10:29just for citizens, but crucially for citizens, for consumers, for
- 10:32creatives. But also, it's abundantly clear
- 10:36that right size regulation is good for innovator, good for
- 10:40investor because of that clarity, certainty, consistency,
- 10:44stability that comes from that. If you wanted a huge
- 10:49underscoring of that point, just look to English common law used
- 10:54around the world. Why?
- 10:56Why used in countries? Why used in agreements where
- 10:58people have no connection to the United Kingdom, have no
- 11:01intention of coming to the United Kingdom, but they still
- 11:04structure their agreements on English common law because of
- 11:09the stability that that brings, because of the certainty,
- 11:12because of the reliability of that.
- 11:14So it clearly makes sense to have the right size regulatory
- 11:18framework around these systems. If we look at the particular
- 11:23model in the state, it's very clear that you can take that
- 11:29approach in an economy in a market of that size, which is
- 11:35pretty much take as much economic resource as you can,
- 11:40take as much compute power as you can, take as much energy as
- 11:45you can, take as much data as you can on one side of the
- 11:48Ledger, Put that into the mix and then see what the outcomes
- 11:53are. It's an approach.
- 11:56The results have certainly been more than interesting, but it's
- 12:00not the approach. There's a whole series of
- 12:03different approaches looking at what you made, used to do with
- 12:08that data before it even goes in to the algorithm, there's
- 12:12looking at how much of it's put in, what purpose are you getting
- 12:16at? So, Sir, sure, it's it's one
- 12:19model, but the claim that it is exceptional, that it is like
- 12:27nothing that's come before in our societies, in our economies,
- 12:32in our democracy, so they need to be left alone, unshackled,
- 12:40unfettered. I completely agree with Chris.
- 12:43And what we're seeing is incredible overdependence
- 12:47beginning to arise. I mean, you, you only have to
- 12:50look at Nvidia's market cap and some of the market cap and some
- 12:54of the other big players. I mean, 5 trillion is a bigger
- 12:58sum than the GDP of most countries apart from the US and
- 13:02China. So, you know, we're talking
- 13:04extremely large entities here. And that dependence on those
- 13:09players, it risks a huge bust at the end of the day.
- 13:15You know, people have been talking about kind of marquee
- 13:17bubble, market bubble. You know, the bankers are
- 13:23already getting pretty nervous. You know, you talked about the
- 13:27investment, Michael. And of course, that kind of
- 13:30vertical integration in terms of data centres, compute power and
- 13:33so on is exactly what the big tech companies are doing.
- 13:39And they're deliberately designing it.
- 13:41So the stack, if you like the, the tech stack is something that
- 13:45isn't really very operable. You're once you start developing
- 13:49your AI model, you're in their stack and it's not that
- 13:54interoperable in terms of being able to ship to any other
- 13:56supplier. So that itself is monopolistic
- 13:59behaviour and you know, will lead to just two or three major
- 14:04large language models being available to the public.
- 14:07So choice will be limited and that will have a knock on effect
- 14:11to governments. Already the UK government really
- 14:15only buys from US vendors. The major tech companies like
- 14:22Microsoft, Google, Amazon and so on.
- 14:25UK developers get don't get a look in.
- 14:28So this has major impacts on economies beyond the US and
- 14:36societies beyond the US. That that concentration risk Tim
- 14:40raises is absolutely critical. There's this be that the
- 14:45economic concentration risk, be it the technology concentration
- 14:50risk. And you only have to look at the
- 14:53AWS outage last week, which was just a minor glitch.
- 14:57You know, these things happen, but look how that rippled right
- 15:03around the world because of that concentration in the small
- 15:07number of players. It's critical for businesses,
- 15:12for governments to really focus on that.
- 15:14Tim Reilly raises concentration risk, economic technology.
- 15:19It really it's, it's a, it's a very, if you're going to start
- 15:21somewhere, it's not a bad place to start when you're making your
- 15:24considerations around these issues.
- 15:26We had a Microsoft glitch. So, you know, I think that's a
- 15:30very good demonstration of that kind of over dependence.
- 15:33We have a a very interesting question from Anthony
- 15:37Scrifignano, our our mutual friend Tim, and he says your
- 15:46points about big tech influencing how we see the world
- 15:50is spot on. Isn't part of How We Win related
- 15:54to our ability to teach the workforce and the user base to
- 15:57ask critical questions about why they should believe what they're
- 16:03seeing from AI? Which of course gets right to
- 16:07the heart of misinformation AI, automated misinformation and
- 16:13disinformation, and the fact that you have programs now like
- 16:16Sora that creates videos that are absolutely fake and look
- 16:21absolutely real. Chris and I are both passionate
- 16:24about media, digital literacy and you know, without that we
- 16:28are even more vulnerable. But if we've at least if we have
- 16:34some media digital literacy, which really government and
- 16:38local government, central government educators, you know,
- 16:42a whole range of people need to get involved in regulators as
- 16:47well. And to make sure that we do
- 16:49develop those critical thinking skills in the face of digital or
- 16:53media, in the face of AI tools. Because, you know, especially
- 16:57with hallucinations from AI models, we need to know whether
- 17:01we're being lied to effectively and whether or not the the
- 17:05information we're getting is correct.
- 17:07You know, we might open a major mainstream media newspaper, for
- 17:12instance, whether digitally or, you know, in the analogue form.
- 17:17But that is going to be, you know, we're pretty sure if we
- 17:20buy the right newspaper, it's going to be accurate.
- 17:23We're not sure about AI. So we need a different set of
- 17:26skills and we need to actually discern misinformation,
- 17:30disinformation, deep fakes. You mentioned Sora Michael.
- 17:34Absolutely. I mean, and so the risks build
- 17:37up. And in the face of that, we have
- 17:39to be, you know, we have to make sure we've got strategies for
- 17:43developing digital media literacy.
- 17:45And I'm sad to say that I don't think in the West, apart from
- 17:48one or two notable countries like Finland and Estonia, we're
- 17:52really are we getting it right yet?
- 17:54Put the human at the heart of all this and to human lead on
- 17:59these technologies, which means imbuing these technologies with
- 18:03our human values and our principles.
- 18:06Trust and transparency, inclusion and innovation,
- 18:08interoperability and international outlook,
- 18:10accountability, assurance, accessibility.
- 18:13All good to be in the mix, but there could never be a more
- 18:17critical time to have all of our critical senses and enabling
- 18:24young people through schooling and through lifelong education
- 18:27to have their critical faculties to the fore.
- 18:31And we've got a Children's Welfare and Schools Bill going
- 18:35through the House of Lords at the moment.
- 18:37I've got a number of amendments down there around these areas of
- 18:41having not only this understanding of how Ed tech is
- 18:44working in the classroom, but to have hard wired through the
- 18:49curriculum. Not just in compute subjects,
- 18:52not just in sciences subjects, but hard wired through the whole
- 18:56curriculum. This sense of digital literacy,
- 18:59media literacy, AI literacy, data literacy, resilience,
- 19:03character education, these are what all of our young people
- 19:07need because that is what will enable them to have their sword
- 19:11and their shield and will be a key part of enabling them to
- 19:14succeed in this digital future now.
- 19:17But how do you go about accomplishing that?
- 19:21You know, Google has search data, Meta has social data,
- 19:24Microsoft has productivity data. How do, how do you, Chris, go
- 19:32about injecting that human inclusive element into these
- 19:41companies that are relying on these enormous concentrated data
- 19:45sets? How, How?
- 19:46How do you do that? First and foremost, because we
- 19:49have great good fortune of still having the most splendid
- 19:57stunning of supercomputers, the most efficient, the most high
- 20:03tech, if you will, human brain, which gives us all of the
- 20:08opportunities that we need. If we deploy it individually and
- 20:14collectively as to what we want, how we want to bring in these
- 20:19frameworks into legislation, into regulation.
- 20:23And none of this means the heavy hammer of bans or controls In
- 20:27that sense. It means having the right sized
- 20:31regulatory approach because everybody flourishes if you have
- 20:36that. And it means nothing short of
- 20:39just having those human values running through all this.
- 20:42Because the truth of it is, I do believe this.
- 20:45I mean, maybe I have a touch of Panglossian naivety about
- 20:51myself, but I think you only as an individual, as a company, as
- 20:57a society, you only succeed long term, firstly together, and
- 21:02secondly, if you have those enduring human values wired in
- 21:07these companies that we're discussing, on one level they're
- 21:10doing extraordinarily well at the moment.
- 21:15But look at 1900. As to how many businesses were
- 21:20on the stock exchanges then to now, look at 1950.
- 21:25Look at 1970. That's not to say we should just
- 21:28wait and see. It means we should be very
- 21:33forward facing, enabling our education system.
- 21:37We've got a curriculum review which is due to report any time
- 21:39in the UK that needs to have all of this wired through it all.
- 21:43But we can lead on this and the worst thing we can do is to
- 21:47think it's all too big, it's all too complex.
- 21:50We retreat from the public sphere.
- 21:53We need to to be even more present, even more connected,
- 21:57properly connected, humanly connected.
- 22:00And we truly then can be very cognizant of the risks and put
- 22:05in the right frameworks there, but really put everything in
- 22:07place to drive, to enable, to empower the upside.
- 22:12Because in no doubt what we can do with these human LED
- 22:17technologies in our human hands. We haven't been able to do
- 22:21anything like that even a few years ago.
- 22:24What we have to be very careful of is thinking of Gen.
- 22:27Z and Gen. Alpha and so on as being digital
- 22:30natives. I think it's actually quite
- 22:34interesting how cavalier young people can be about their own
- 22:40data. And you know, Chris is right, we
- 22:43need to have that in, you know, all our institutions.
- 22:46One of the big issues we have is with another generation, our
- 22:49teachers, you know, upskilling them is an absolute priority in
- 22:55terms of understanding how they can then impart, if you like,
- 22:59digital safety messages and critical thinking messages to
- 23:04the, the, the young people that they teach.
- 23:06But Chris is right, you can't do without regulation.
- 23:09And you know, we're obviously experimenting with our online
- 23:12safety legislation. There's been huge push back from
- 23:16the tech companies. We're age gating at 18.
- 23:21We're trying to make our Internet safe for children in
- 23:25particular, but we are, you know, facing some really big
- 23:30issues because how do you capture all the harms that are
- 23:34occurring online? And new harms are occurring all
- 23:38the time. You know, for instance, we don't
- 23:40think we captured the use of chat bots online enough in
- 23:45social media directed at children.
- 23:48There are things like live streaming which we don't think
- 23:52are fully captured properly. You know, there are to be agile
- 23:56as a regulator is very, very difficult when you're
- 23:59interpreting legislation that may not have caught up with the
- 24:03new harms as they occur. And that is something that as
- 24:07legislators, you know, Chris and I have to work with.
- 24:11We're slightly impatient of regulators, but sometimes if the
- 24:14regulator isn't at fault, it's us.
- 24:16As legislators, we didn't encompass what we really needed
- 24:20to encompass with the legislation and we we pay the
- 24:23price for that. We have a bunch of questions
- 24:27that are coming in. Let's let's jump to some
- 24:30questions. But I just have to respond to
- 24:33one comment that Chris made. He referred to himself as
- 24:38Panglossian, as potentially having Panglossian naivete.
- 24:43And I was thinking, Chris, as you were talking that you're
- 24:46really an optimist. I'm a rational optimist, yeah.
- 24:50OK, let's jump to. I'm a conditional optimist.
- 24:54So we have a rational optimist and a conditional optimist.
- 24:58I'm an optimist as long as there's regulation down the
- 25:01track, you know? Well, what are you, Michael?
- 25:05I am a realistic optimist. I'm not sure how that's any
- 25:10different from either of you, but or a sceptical optimist,
- 25:15because I've worked with so many tech companies over these last
- 25:19decades and I understand the motivations and I understand the
- 25:23pressures and the goals and the power of these companies and the
- 25:28level of resources. I mean, the watchword for me is
- 25:31diversity in so many ways that I, you know, I have a, I'm a
- 25:35great believer in a diverse society because you get a range
- 25:39of ideas, you get a range of creativity.
- 25:42No, it isn't one-size-fits-all. It isn't conformity.
- 25:46And if you find yourself focused on only three or four major
- 25:52businesses in any particular sector, you're getting a
- 25:56conformity. You're getting, you know, a lack
- 25:59of competition and a lack of diversity.
- 26:02I once did an analysis of of major software companies who
- 26:07were producing ERP products and I did a word analysis and it
- 26:14turned out that across all these major companies, the, the
- 26:18marketing, the messaging was virtually identical because
- 26:21there's so much movement among the leaders of these companies.
- 26:27That's been so interesting. You see Microsoft acquiring, you
- 26:31know, the, the, the ex CEO of DeepMind Googled, you know, and
- 26:37so on. And the shifting between those
- 26:40big tech companies is is significant.
- 26:43If you're watching, this would be an excellent time to
- 26:46subscribe to the CXO Talk newsletter.
- 26:50Join our community. We will notify you of upcoming
- 26:53shows just like this. Go to cxotalk.com and sign up.
- 26:58Just take a second. Do it right now.
- 27:01All right, we were on Twitter and now let's take some
- 27:05questions from LinkedIn. And Jerry Kopitch has been
- 27:09waiting patiently and he asks this question.
- 27:12He says How insulated is the UK if the global AI economic bubble
- 27:20bursts? We're pretty vulnerable
- 27:23actually, especially our UK government because you know,
- 27:27and, and Chris mentioned the, the AWS outage recently, we had
- 27:35a Microsoft outage as well. These, this is, you know,
- 27:39vulnerability on a big scale if the whole of your government
- 27:43depends on three or four US supplies.
- 27:46I mean, if Palantir went down for any particular reason, our
- 27:50government in defence and across our health service, you know,
- 27:55could well be in trouble because Palantir helps to organize all
- 27:58the data within those two services.
- 28:01So, you know, we haven't mentioned them, but they are a
- 28:04significant player internationally in those two
- 28:08sectors and in the UK as well. So I, I, you know, this is, this
- 28:13is no small thing in terms of, you know, what could happen in
- 28:17the future and, you know, cybersecurity, if we've seen
- 28:21vulnerabilities on a major scale from companies like Microsoft in
- 28:27the past. And so again, if you only have,
- 28:31you know, one or two major suppliers, whether it's
- 28:34platforms for, you know, operating systems or, or, or
- 28:39mobile platforms, you're again, you know, creating
- 28:43vulnerabilities for yourself. Once bad actors find a way in,
- 28:49then whole businesses are down. We recently had a major outage
- 28:54for our biggest car manufacturer for Jaguar Land Rover and
- 28:59they've been down for months and months and that the cost of the
- 29:03business is billions, a huge hole in their bottom line going
- 29:09forward. Now luckily, you know, they're
- 29:12owned by, you know, a major in Indian company and you know,
- 29:16they can just about afford to go through it.
- 29:18But their suppliers, parts and so on in the supply chain have
- 29:23been really badly affected and had to be propped up by loans in
- 29:29the UK government and so on. So you know this is not some
- 29:32theoretical dependency. This is a huge issue.
- 29:37Chris, we have a question from Arsalan Khan.
- 29:39Let me direct this to you. And I think it's quite related
- 29:43to this set of issues. Arsalan Khan says on Twitter
- 29:48companies are blaming AI for their massive layoffs.
- 29:53How can or should government even excuse me?
- 29:57How can or should government even step in in this rapid
- 30:02change? What can government do that
- 30:05doesn't take 10 years? It's a really key point because
- 30:08it takes us back to the human in all of this.
- 30:12And if the human must be at the heart of the technologies, and
- 30:17if the human must be leading on these technologies, then the
- 30:21human in terms of quality, meaningful work needs to be at
- 30:29the heart of all this. The key issue with all of this,
- 30:33I don't think is considered enough, is how you lead on the
- 30:37transition. I think that as with other huge
- 30:44shifts in our time, that AI and other technologies will be job
- 30:51creating. And that's ultimately fine, But
- 30:56the transition as it has been when other industries have
- 31:00started to wane and decline, you have to as a government, you
- 31:05have to society understand that transition and how you're going
- 31:09to manage through it and lead through it.
- 31:12Because nobody is going to accept huge layoffs with no
- 31:18societal, no governmental level consideration on this.
- 31:23And we've seen it with other industries when they have
- 31:26started to disappear. That hasn't been fully
- 31:28understood. It hasn't been properly dead on.
- 31:29It hasn't there hasn't been the public debate, the discourse,
- 31:32the how do we all go through this together?
- 31:35We can't just have, for example, many businesses not taking
- 31:42graduates newly qualified on. And that's already happening
- 31:46massively across all of those sectors.
- 31:48Well, how as a young person do you get your career underway?
- 31:51How do you get to be that sort of for five years into your
- 31:55career if a lot of the entry level work and tasks is being
- 32:00taken over by AI? Well, there is absolutely a way
- 32:05through that and that is focusing on the individual as
- 32:09human aptitude, as human skills. But it requires leadership, it
- 32:14requires a re imagination of how work is structured, how job
- 32:20roles are structured, understood, advertised for and
- 32:26on boarded for. Because the reality is AI isn't
- 32:31as such wiping out jobs. It's doing things.
- 32:35It's doing particular tasks, it's doing particular part of
- 32:38particular role. Critically important to
- 32:39understand that. But that is the start point to
- 32:42then understanding how to structure how to lead on the
- 32:47labour market for the future, where, as Tim mentioned a few
- 32:52minutes ago, the right way to do it is to focus on that diversity
- 32:57and that inclusion and to lead on that.
- 33:00And then we should be rationally positive, rationally optimistic
- 33:04as to how we can go about it. We do have a paradigm for this,
- 33:07of course, in terms of how the car, the the horse gave way to
- 33:11the car. And I, you know, often think
- 33:14about that photograph of 5th Ave. in 19 O3 where there was 1
- 33:19horseless carriage driving down 5th Ave.
- 33:23All the rest were traps, pony traps, horse traps.
- 33:29And then ten years later in 1913, you didn't see a single
- 33:32horse and carriage going down 5th Ave. in that famous photo.
- 33:36And what we have to, we don't know yet what jobs are going to
- 33:40be there in 10 years time. And so we have to worry about
- 33:45this lack of entry level jobs taking place.
- 33:49And we have to be absolutely intent on upskilling in the way
- 33:54that Chris described. And not just for the younger
- 33:57generation trying to get jobs at the moment.
- 34:00We have to be upskilling, you know, the older generation as
- 34:04well because jobs are being substituted or even in big tech.
- 34:08I mean, Microsoft is shedding thousands of jobs currently as
- 34:12we speak. And many of the other big tech
- 34:14companies are finding that they can do their coding through AI,
- 34:18not through using humans and so on.
- 34:22And so, you know, they're noticing that and other
- 34:25businesses, which may be even more, be more analogue, is still
- 34:29going to have to shed jobs in order to remain competitive.
- 34:35But it's because we don't absolutely know what new jobs
- 34:38are being created. We have to be very agile in all
- 34:41of this. And you know, we've been here
- 34:43before and it wasn't a happy time.
- 34:46What is the role of the government in this?
- 34:50Are you going, are you advocating that the government
- 34:53forces companies to to undertake this reskilling?
- 35:01Is it government programs? What's the right construction?
- 35:04I think so they have to be more activist and there are some good
- 35:07models out there. But, you know, recently
- 35:10governments have adopted a kind of lifelong learning approach,
- 35:14but without really giving it enough teeth and a resource in
- 35:17order to do that. They've accepted the obligation
- 35:20without really giving the tools to those who are trying to
- 35:25deliver that lifelong skill set. If you like.
- 35:28So you can retrain if your job becomes, you know, out of date
- 35:33in terms of the tech technology, you should be able to upskill
- 35:38with a grant or with a loan or whatever.
- 35:40Now there that is coming into place, but it's not nearly fast
- 35:44enough. And for young people, you know,
- 35:46apprenticeships, particularly in some of the industries where
- 35:49it's freelance, freelance working is the norm.
- 35:53Again, you know, there isn't enough apprenticeship going on.
- 35:57And the kinds of taxation that we have, it doesn't help with
- 36:01that. So you know, we haven't yet got
- 36:03our skills right, but it is something that government should
- 36:07be doing and also obliging larger companies to undertake.
- 36:13In schools have it threaded rights through the curriculum
- 36:16and indeed further and higher education.
- 36:18And then when we get into your career and the workplace itself,
- 36:22bite sized badges, boot camps, all of those ends where you can
- 36:28at speed in an agile manner, get these skills, get these
- 36:32competencies while you're in role.
- 36:36So you have speed, you have pace, you have agility as you're
- 36:40moving through your career. It's doable.
- 36:42But ultimately, as so much, it comes down to leadership, and
- 36:46that leadership should come very much from the top.
- 36:49When you say come from the top, you're referring you from the
- 36:52companies or from the government.
- 36:54It should come right from the Prime Minister and then every
- 36:57level down and then in companies, in all businesses, in
- 37:02all organizations, as a leadership role for everybody to
- 37:05take at their level of the organization to lead on this.
- 37:07So from your perspective, this is AI don't want to put words in
- 37:11your mouth. It seems a a major fundamental
- 37:17society mandate if we put put it that way.
- 37:22Last year, 2 billion of us went to the polls and democracies
- 37:25around the world. And yet, where was any of this
- 37:29talk in those manifestos, in those party election broadcasts,
- 37:36in that election literature, where was anything talking about
- 37:39what could AI, what could blockchain mean for you, your
- 37:45families, your businesses, your communities, your cities, your
- 37:48country? And yet if we look at the major
- 37:51challenges of our time, climate crisis, energy emergency,
- 37:56migration, cost of living, we're all unbelievably connected and
- 38:02yet there's a loneliness epidemic.
- 38:04Well, technologies have a lot to say for all of those issues if
- 38:10they are human LED technologies in our human hands.
- 38:14What phenomenal manifestos parties could have written
- 38:18talking to all of those issues of our time.
- 38:21So many global in nature, so many existential in nature, and
- 38:25yet human LED narrative, human LED technologies enabling that
- 38:31narrative that could have been so positive to draw people into
- 38:34the public discussion, the public discourse around how we
- 38:38go about these new technologies. And yet, frankly, to find much
- 38:43mention in any of those, you'd have needed to find my great,
- 38:48great great grandfather Sherlock Holmes to try and be deployed to
- 38:52try and find much mention of this.
- 38:54I mean, people would think of the first duty of governments
- 38:56being security, if you like, but livelihoods are part of that
- 39:01security in my view. And you know, pathways to jobs,
- 39:06education and so on must be a vital role for government.
- 39:10I don't think that governments can just simply delegate that to
- 39:15business. This is a question from Brian
- 39:18Mosley. And he says how do we protect
- 39:21ourselves from the impending weapons of mass persuasion?
- 39:27And he says by Elon Musk and Mark Zuckerberg.
- 39:30But let's just make this weapons of mass persuasion, meaning AI
- 39:35generated misinformation and disinformation.
- 39:39Back to critical thinking, and also I think how data is
- 39:43directed to you in a targeted kind of a way.
- 39:47We all have to be increasingly careful about the use of our
- 39:50data. The kind of surveillance
- 39:52capitalism that Shoshanna Zubov talked about, where, you know,
- 39:56basically you're giving your data to social media and then
- 40:01social media is targeting you with advertising and so on.
- 40:05Well, increasingly that's going to happen in terms of
- 40:07misinformation, disinformation. And already is, quite frankly,
- 40:12you know, for instance, you know, messages during election
- 40:15campaigns and so on, and how much regulation is there of
- 40:19those kinds of messages in most of our democracies?
- 40:23So we're already, you know, in a, in a real, real problem.
- 40:28And we have to, again, we can't just rely on regulation.
- 40:32We have to rely on our own critical faculties.
- 40:35Critical thinking, Critical senses to the fore.
- 40:39Test, test, verify. Take that second, that third,
- 40:43that fourth click and always be so present and so alert.
- 40:48Because the government have a key role to play.
- 40:51We have to be absolutely tooled up ourselves as citizens, our
- 40:56swords and our Shields to the fore.
- 40:58Here's a question from doctor Carolina Sanchez Hernandez, who
- 41:05says, what do we do when no one is able to explain the outputs
- 41:11of AI, when a person is unable to get a loan or insurance?
- 41:17Because these are black box systems, very often the
- 41:20algorithms. Timmy wrote a whole book about
- 41:22this. The big challenge is oversight
- 41:25going forward and also oversight of the entire AI value chain.
- 41:31So what do we do about this black box element with that's
- 41:34making key decisions over our lives?
- 41:37Chris and I are passionate about a regulation that makes sure
- 41:42that you do get that element of transparency and you do get
- 41:45redress if decisions are made by this black box and you're not
- 41:51informed about it or it goes it's biased, it, it, you know,
- 41:55there's, there isn't a human in the loop and so on.
- 41:58And, you know, there's no escaping it if we're going not
- 42:02to be completely subject to AI decisions, whether by government
- 42:07or private businesses like insurance or bankers or whatever
- 42:11it may be. You know, we need to have that
- 42:13level of regulation. You know, you can't just rely on
- 42:16voluntary agreements. And, you know, we talked about
- 42:19being tech optimists or tech realists or whatever.
- 42:26Well, you know, I'm a, I'm an optimist, but it's conditional
- 42:30on making sure that we have this kind of regulation in place.
- 42:33Because otherwise all we're doing is making sure that people
- 42:38are living in a world which is completely opaque, where
- 42:41decisions are made about them. They've got no control over and
- 42:44no opportunity of redress. So that's not a world I, I want
- 42:48to see. It's where literally, you know
- 42:51the machine AI is our master, not our servant.
- 42:55To quote the title of of my book.
- 42:58It was a great advert there for Tim's book and I recommend it to
- 43:01everybody. And he hasn't even offered me a
- 43:03percentage for that endorsement of it, but it is a good read.
- 43:07So if you haven't, do get your hands on it.
- 43:10And similarly, I wrote a report on this subject earlier this
- 43:15year and I entitled it 8 Realities, 8 Billion Reasons to
- 43:21Regulate. And I just set out exactly what
- 43:23this question gets to the person who doesn't get a loan and
- 43:28doesn't even know that AI was in the mix, never mind that AI was
- 43:33the rejector of their loan application.
- 43:36I looked at the scammed, I looked at the teacher, I looked
- 43:39at the transplant patients, people on the liver transplant
- 43:42list in this country. AI is involved there.
- 43:45Do they even know that's the case?
- 43:47So First things first, know that AI is there.
- 43:52So labelling transparency, because without those two things
- 43:56how can you possibly have anything which looks like trust?
- 43:59Secondly, do not for one minute accept this argument that it's a
- 44:03black box. It's just too complicated.
- 44:06We can't know what's going on there because when the
- 44:09organisations need to know, you'll find out that it is
- 44:13possible to know what's going on and you can make those changes.
- 44:17So the legislation I brought in, in draft in 2023 was too many of
- 44:23these points to look at the right regulatory framework for
- 44:28these technologies which we call AI.
- 44:32And I think as a formulation, if we are principles based,
- 44:37outcomes focused and inputs understood, that really puts us
- 44:43the right framework to go about this.
- 44:45And as you see that outcomes focused inputs understood, you
- 44:49can see by doing that you really squeeze that supposed black box
- 44:54in the middle. Let's jump to another question.
- 45:00This is from Huawei Wong on LinkedIn and she says and she's
- 45:06a she's a senior data scientist. So this is from that
- 45:09perspective. And, she says, what prevents
- 45:14innovators from being reduced to data tenants only in the Big
- 45:20Tech's empire? I'm passionate about the open
- 45:25source movement myself. And I think a way of escaping
- 45:30from just being, if you like the slaves of of the big language
- 45:35models in that sense is if you've, if there are open source
- 45:41models which you can piggyback on and create new forms of AI
- 45:49tools using that those bigger models.
- 45:54I mean, you know, I'm not a huge fan of the of the Lama large
- 46:00language model from Meta, but it is open source and it is
- 46:04available by and large for people to experiment with, build
- 46:09their own models and so on. That is not true of Claude or
- 46:15ChatGPT in quite the same sort of way.
- 46:18So, you know, there are the possibilities of being able to
- 46:21use the power of models that have been created and that will
- 46:28then create a diversity of AI tools that that would be
- 46:32available. And I think that's really what
- 46:34we look to in the UKI don't think we're going to have the,
- 46:38the kind of compute capacity, the sheer sophistication of
- 46:43chips and so on, which, and indeed even the financial
- 46:47resources to create our own large language models in quite
- 46:51the same way. So we're going to want to
- 46:53piggyback on some of those. And I think that's the answer.
- 46:56The, the data sets that we'd be using would be not quite as
- 47:00comprehensive and scraped from every bit of the Internet in the
- 47:03way that they are in the, the, the clawed and, and ChatGPT.
- 47:08They would be more proprietary. And probably we, you know,
- 47:11hopefully people will have paid for the rights to access the
- 47:15data so that then there'd be benefit for those who created
- 47:18those data sets. So again, I think that would be
- 47:22beneficial in many other ways as well, rather than the current
- 47:26situation where people like Open AI and Anthropic pretty much are
- 47:31denying that they need to pay anybody for training their large
- 47:34language models on what is often copyrighted.
- 47:37I. Think open source has the
- 47:41potential. It's not inevitable, but open
- 47:43source has the potential to be the special source in this and
- 47:49building on that, to understand that the approach and the models
- 47:57that are built and deployed out of the valley, That's a
- 48:03particular approach, but it's one approach.
- 48:07There is a whole array of different ways of going about
- 48:11this, a whole array of innovative ways of going about
- 48:16this. If we consider where certain
- 48:18data rests, if we consider some of the data sets that we have in
- 48:22this country, if we really understand all of the issues
- 48:26around privacy, if we understand all the issues around whose data
- 48:30it is, who owns it, who has the rights to that data.
- 48:35If we start to talk much more about the potential for the
- 48:39National Data Library, the NDL on what could come about, there
- 48:43many, many reasons to be rationally positive about this.
- 48:49But we will get this fundamentally wrong in the UK if
- 48:54we believe that the US model is the only model or we seek to
- 48:59just simply copy or eight that model because we will just be
- 49:04Coke light if we do that. And we've got our debate raging
- 49:09in the UK, and I know that there are many cases in the US,
- 49:14Michael, where copyright owners are suing, you know, the big
- 49:19tech companies for infringement by training them, you know,
- 49:24material. They're training their models on
- 49:26copyright material. And all those cases are yet to
- 49:30come through, you know, to see what the final decision is.
- 49:33But certainly, our government has had a lot of criticism for
- 49:37trying to give a free pass to big tech companies in their
- 49:42training. And it's going to continue to be
- 49:44extremely controversial even when we come to new legislation
- 49:48next year because we're due to have an AI I Act AAI bill come
- 49:52through next year. And of course, speaking for
- 49:56everybody in the US, we here in the US do believe that our way.
- 50:02I'm not going to say it's the only way, but it's the right way
- 50:06for everybody else. OK, moving on.
- 50:09Very. Yes, I mean, but the, the, of
- 50:11course, the approach to data is very different in the states.
- 50:14And what we're thinking of more is how individuals can in a
- 50:19sense collect the data with their in within communities, if
- 50:23you like, but for public benefit.
- 50:25And but it does mean that they will have greater control over
- 50:29their own the data that they're creating in their daily lives.
- 50:33So it is a different approach. AI infrastructure is becoming as
- 50:38necessary as electricity. Given this, should data centres
- 50:43be treated as regulated monopolies?
- 50:47I think there's a question which is worth debating around a
- 50:51number of elements of these technologies as to what's the
- 50:53role of the state, what's the role of the business?
- 50:56A word Tim used a long while ago is absolutely right.
- 51:00This sense of getting the right partnerships will get us to the
- 51:03right place. I think they will be regarded as
- 51:06a utility ultimately, like gas or electricity or whatever,
- 51:10because they're such an important part of the
- 51:12infrastructure, quite honestly. And also the cost of developing
- 51:17these data centres is becoming so astronomically expensive.
- 51:24Yes, the environmental cost included.
- 51:27Yes. And there's also the impact on
- 51:30electricity prices. I mean, here in the US, our
- 51:33electricity prices have significantly increased over the
- 51:36last year or two. You don't know anything compared
- 51:39to the UK, Michael. Seriously.
- 51:42And it's all jammed. Tomorrow we're told that small
- 51:45nuclear reactors are going to be the solution, but they won't
- 51:49come on stream for, you know, a decade or so.
- 51:53So you know, this is this is a bit pie in the sky at the
- 51:56moment. Chris, I'll have to mention what
- 51:58you said about what's going on in the UK to my wife, who
- 52:01recently complained to me that our electricity bill had
- 52:03doubled. Well honestly, you are.
- 52:06You are still very much tucking into a nice pot of jam compared
- 52:09to the UK situation, I promise you.
- 52:12I think you better come on holiday here and just see see
- 52:15what it costs in your holiday cottage, Michael.
- 52:18How do we strike the right balance between regulation for
- 52:22the protection of society versus protecting new technologies?
- 52:27What we have to do is understand that this kind of mantra, that
- 52:31regulation is the enemy of innovation, is not the way
- 52:35forward. And what we have to do is
- 52:37understand that the proper forms of regulation involve assessing
- 52:41the risk, making sure that the standards you're imposing create
- 52:46an interoperability with other the countries.
- 52:49You know, we're not sitting there in isolation.
- 52:52So you're not creating a kind of Oasis for your own developers
- 52:57who can't then get out and sell their wares and sell their
- 53:01products overseas. You've got to make sure that
- 53:04you, you know, you're, you're, you're, you're regulating in an
- 53:07appropriate fashion. And I would say that, you know,
- 53:10you've got to have the risk assessment.
- 53:12If you've got high risk AI being created, you've got to make sure
- 53:17that that has certain duties of transparency, accountability,
- 53:21liability and so on and that the consumer, if it's a consumer
- 53:26facing has the right of redress. If it's business, you know, they
- 53:31have other rights relative to that.
- 53:34It's this is not some, you know, a terrible idea.
- 53:41We regulate so many areas. Aviation, the motor car wouldn't
- 53:45have taken off unless we'd had safety regulation for, for the
- 53:49car, you know, a consumer products, most of them have
- 53:53safety regulation and that's what we need for, for AI.
- 53:57Open AI has lost billions. It survives only because of
- 54:01investment from companies like Microsoft.
- 54:04Anthropic survives because of investment from Amazon.
- 54:09Can we make the argument that market concentration is a great
- 54:14thing because it's enabling AI? And you can make the argument
- 54:19that it's a great thing until it isn't.
- 54:21It's something that a lot of us should be at least somewhat
- 54:26concerned about, that economic concentration and the
- 54:30circularity of funding and applying all the principles that
- 54:34we know. So looking as to what you're
- 54:36investing in, what is the return on investment, applying all of
- 54:41those good economic and business principles that are well made
- 54:45will give us the right answers and the right analysis.
- 54:48And and of course, this is going to be 1 great bubble at the end
- 54:51of the day, I think we're all a little bit apprehensive about
- 54:54this. Chris brought up this issue
- 54:56which we have not touched on, which is the circularity of
- 55:01funding, just your views on that.
- 55:04It's such an important issue and we were, but we're out of time.
- 55:07Yes, absolutely. And this is the exactly why
- 55:10people are worried about this bubble because, you know,
- 55:12Microsoft is investing in open AI.
- 55:16Open AI is buying credits from Microsoft.
- 55:19And you know, NVIDIA is doing pretty much the same.
- 55:23And so they're feeding each other economically.
- 55:26And if confidence goes, and if open AI doesn't make any money
- 55:32and isn't making a great deal of money at the moment, then what
- 55:35happens? You know everything goes goes
- 55:37bust and you know it. It all depends on Microsoft
- 55:40continuing with its mainstream business rather than with the
- 55:44the large language model ChatGPT.
- 55:47What should business leaders and boards of directors do in this
- 55:51complex environment with shifting regulations, changing
- 55:55technologies? Your advice please for business
- 55:57leaders and boards. Get involved, get up skilled,
- 56:01get knowledgeable, get involved in the debate and discussion,
- 56:05understand what you're trying to achieve as a business,
- 56:07understand how potentially these technologies could help and
- 56:12partner up and really connect and make it a team effort.
- 56:16Know what your values are in the middle of all of this, because
- 56:20there's nothing like AI for making you question your values
- 56:23in all of this. Know what your values are.
- 56:28That's an abstract concept, potentially.
- 56:31It is potentially abstract, but frankly, what is the purpose of
- 56:35your business? You know, well, how do you treat
- 56:38your employees? You know, what kind of trust do
- 56:42your consumers do, your employees and your other
- 56:45stakeholders having you? Well, often that's all about the
- 56:48values that you as a business have.
- 56:51And I think we forget about them too often.
- 56:54Chris, what advice do you have for regulators and policy makers
- 56:59when it comes to the set of AI issues?
- 57:02Understand what your role is, understand how these
- 57:07technologies are developing and how they're already being
- 57:10deployed and look about an approach where you do right size
- 57:15regulation and you understand the complexity of the needs of
- 57:19the consumer, the creative, the citizen, the investor, the
- 57:23innovator. It's not straightforward, but
- 57:26it's completely doable. And we've seen examples in other
- 57:28areas in the UK where we've done it.
- 57:30So we know how to do this, we need to do it.
- 57:33And in the words of a Great American company, their best ad
- 57:36campaign ever, Nike, ultimately, having done all that, just do
- 57:41it. Tim, you're going to get the
- 57:43last word advice for regulators and policy makers.
- 57:47Resist the blandishments of Big Tech, who will tell you that,
- 57:51you know, all regulation is terrible, they won't be able to
- 57:54innovate and do business, but you have to resist that.
- 57:58We know the resources that are going into this from Big Tech,
- 58:03both in Brussels, London, Washington and so on, and we
- 58:07need to be resistant to that. How do you resist?
- 58:11They have so much money and resources to throw around and
- 58:15share. How do you resist?
- 58:17You resist by coming on CXO talk and influencing the influences,
- 58:24Michael. That works for me.
- 58:27And with that, a huge thank you to Lord Tim Clement Jones and to
- 58:32Lord Chris Holmes. I am so grateful to you both.
- 58:34Thank you for taking your time and sharing your expertise with
- 58:38us once again. Thank you very much.
- 58:42Pleasure as always. And I hope you'll both come
- 58:44back. Everybody, thank you for
- 58:46watching. Now before you go, subscribe to
- 58:50the CXO Talk newsletter, go to cxotalk.com.
- 58:53We have amazing shows coming up. Check it out.
- 58:58We'll see you again next time every.
- 59:00We'll see you again next time everybody.
- 59:02Have a great day and. We'll.
- 59:06Have fun with AI.