House of Lords Members on AI:
Does Big Tech Own You?

The AI Oligopoly and Big Tech's Influence

Michael Krigsman: Microsoft has invested $13 billion in OpenAI. Amazon poured billions into Anthropic. Google controls your search data. Meta has your social graph. The AI oligopoly is here today on CXOTalk number 899.

Lord Tim Clement-Jones and Lord Chris Holmes, key leaders shaping UK AI policy in the House of Lords, reveal what business leaders must know now. I'm your host, Michael Krigsman. Let's get into it.

Tim, tell us about your work in the House of Lords and your focus areas.

Lord Tim Clement-Jones: I speak for my party on all matters around innovation and technology, and I chair a cross-party group on artificial intelligence as well, which I founded back in 2016, '17.

And one of the key issues for me is the power of Big Tech, which really is increasingly, it's going back in a sense to the big oil issues back in the 19th century, where you had vertical integration of an absolutely essential utility by a limited number of businesses, and that's what we're seeing increasingly.

So already we had a lot of Big Tech concentration. We've got control over two platforms really, Apple, Google. We've got cloud services really in the hands of say two, Microsoft and Amazon Web Services. We've got concentration. We've got Facebook really, which is along with TikTok, one of two really major social media outlets.

So we're getting this concentration in digital, which is now being replicated in artificial intelligence, and especially with the links between OpenAI, for instance, and Microsoft, between Amazon and Anthropic, and of course Meta with its own large language model.

So this kind of concentration is beginning to impact in a major way, which I think will have knock-on effects for society, for our democracy, for countries trying to develop their own AI models.

Opportunities and Risks of AI Development

Michael Krigsman: Chris, tell us about your work in the House of Lords and any initial thoughts that you have on this Big Tech set of issues.

Lord Chris Holmes: I try and cover new technologies, including AI, blockchain, cyber and all things digital assets and tokenization, with threads of ESG and EDI running through all of those. Also cover financial services regulation and all things FinTech.

And really, what I'm trying to bring out are the possibilities, the potential of all of these new technologies of themselves, and then looking at some of the really exciting value cases where you look at the technologies in combination to one another.

But we only win with these technologies if we human-lead on them. We only win if we focus on the public good, the common good, and see them as enablers, empowerers, and really in all our human hands. That's the opportunity, but as we all know about opportunities, they're never inevitabilities. It's down to us to make it so.

Michael Krigsman: Tim just described this concentration of Microsoft and OpenAI, Amazon Anthropic, Google Gemini and DeepMind. You have Meta with LLaMA. What does this concentration mean for competition and for society?

Lord Tim Clement-Jones: It means concentration of information sources, effectively. I mean, one of the key things, this isn't just about people accessing large language models for the hell of it and for use in their daily life. And I entirely agree with Chris. The opportunities are there, but the risk is that what we're doing is we're getting our information from very limited sources.

I mean, for instance, the impact on news media of the digital era has been considerable with the monopoly of advertising that is now pretty much held by Google and Facebook. But of course, with the advent of artificial intelligence, which in a sense scrapes news data from the internet, that means that concentration and that impact on the news media is even higher.

So this is becoming of great importance in terms of how people see the world. And if people grow up and they get all their information from one of the large language models, then they are directly influencing what we're doing and seeing.

And I think that just digressing slightly, but robotics is going to pronounce that even more when AI is embedded in robotics that we use for everyday life. So this is not a small thing.

Lord Chris Holmes: There's no way that we can say that we win with these technologies if we just have a winner takes all reality. There's no question these companies, we all know the list of them you set out at the outset, Michael. They've come an extraordinarily long way in a relatively short space of time. Their business model is clear. Their scaling has been clear. And that is one particular model, one particular approach.

But we win with these technologies if we truly have a plurality, an understanding that AI isn't one technology, AI isn't just LLM, which I think the public narrative is tending towards that sense that we've been on an AI journey and we've now reached this point and this is it, this is the zenith in many ways. Large language models, that's where we've been shooting towards.

That's but one element of AI in a whole constellation of technologies. So I think each individual, each company, each community, each country needs to consider what do they want their AI, what do they want their technology story, their technology narrative to be? How do they play to the elements, the values, the beliefs, the social, the democratic structures in that nation?

And really thinking deeply into this, and that will be the way that we succeed within this. It can't be that we just go down a route of winner takes all.

Lord Tim Clement-Jones: Chris is absolutely right. Increasingly, I think, and this is if we're not careful going to lead to a kind of nationalism as far as AI is concerned, there is this very strong mood both in countries within the European Union, but also in the UK, that we need to be able to develop our own sovereign AI capabilities.

And that if we're not careful, US companies are going to drown out our own ability to develop AI tools of every description. And so this, if we're not careful, this is gonna be a backlash.

And you know, I want to avoid that because I think we need to make sure that we take advantage of every kind of tool that is available internationally. But if we're dominated by three or four businesses, then I think that's going to create a reaction which would be extremely unhelpful.

And we've seen what happens if you start applying sanctions and tariffs and so on. People get rather excited about that, and I don't believe that that is going to help develop our, if you like, technological partnerships across the world.

Michael Krigsman: Folks, we have two tremendous guests today. Ask questions. If you are watching on LinkedIn, pop your question into the LinkedIn comments. If you're watching on Twitter, use the hashtag #CXOTalk.

And we have some questions that are coming in already relating to the international aspects and the economic aspects, and we're gonna get to those questions momentarily. But I have a question for you both.

Regulation and the Future of AI

You've both raised certain types of cautions. You've described the potential very significant impacts on society. You've described in a sense what could be, from a business and from a social perspective, almost a scary scenario.

Lord Chris Holmes: It could be a scary scenario, or it could be an amazing opportunity. It's in all our hands. So we've got to get involved, get upskilled, stay knowledgeable, and really take the work seriously.

And businesses need to consider what is their AI plan, their tech plan. Individuals need to think through what is their engagement with these new technologies. And regulators and policymakers have to absolutely stay on top of this in a way that we actually have the regulatory sandboxes and frameworks to enable innovation but also protect.

And we've got to make sure we've got that right balance, that harmony. Otherwise, yes, it's going down a dystopian pathway.

Michael Krigsman: But Tim, you have a regulatory framework in the UK. The EU has its framework. The question becomes, how do you actually enforce any kind of meaningful regulations?

Lord Tim Clement-Jones: The power of the regulator depends on the legislative base. So for instance, the Competition and Markets Authority in the UK has limited powers. Now, they're trying to use them to the best of their ability, but they're looking at whether something is anticompetitive within the UK context, given the existing powers that they've got.

But they don't, for instance, have powers of merger control over AI like they do over other mergers, and they don't have powers of kind of ex ante regulation of Big Tech generally, not just AI.

And so, the government's looking to increase, if you like, the powers of the CMA to take on some of these issues. But really, you either need a beefed up kind of digital markets act like the European Union does, with teeth in terms of the fines that can be handed down, or what you really need is...

Michael Krigsman: Wait, wait, wait. Hold on, hold on. The EU has a framework, the AI Act and so forth. How effective do you think that will be?

Lord Tim Clement-Jones: Well, I think the AI Act, there's a lot of detail to be worked out. But the key is, you know, how long does that implementation take? The devil is in the detail and the practicality. What does it actually say? The framework's really important. And it's... We also have a very important principle based approach in the UK, but if there's no teeth, in other words, if you don't have serious fines, you don't have serious enforcement, then it doesn't work.

So you need both. You need the framework, the flexibility, but you also need the kind of fines which, like in the European Union with the Digital Markets Act, could be up to 10% of global turnover, which is very, very serious money indeed.

Michael Krigsman: And Chris, what about this issue? Because it seems like there's almost a disconnect between the regulatory framework and the resources and the determination to actually enforce it.

Lord Chris Holmes: Yes, this is absolutely key, Michael. Because you can have all the fine words in the world around regulation, but if you don't have the determination, if you don't have the resources, if you don't have the know-how, if you don't have the skills, if you don't have the upscaled regulators, then nothing. Words on a page.

So we've got to have regulatory frameworks. They've got to have teeth. They've got to be dynamic. They've got to be able to keep pace with the speed at which the technology is moving. But alongside that, we've got to have the regulators who can deliver that, regulators who've got the funding, who've got the people, who've got the capability.

Then and only then do we get meaningful, effective regulation. Otherwise, it's a shield behind which Big Tech can hide, saying, "Oh, we've got regulation, everything's fine," when actually, it's a smokescreen and nothing's actually happened.

Michael Krigsman: Arsalan Khan on LinkedIn asked this question. He said, "I understand the concentration is problematic, but do you think the innovation that has come from Big Tech competition has actually driven us forward?" So in other words, is the concentration of Big Tech actually a feature rather than a bug? That's sort of what Arsalan's getting at.

Lord Chris Holmes: Yes, you've got to say they have come a long way in a very short space of time. They have delivered innovation. They have moved the dial forward. The question is, is that the only story that we want? Is that the only way that we can get there? And is winner takes all the ultimate end reality?

And I'd say that we can have plurality. We can have diversity. We can have competition. And with that, we will see innovation. The innovation to date shouldn't blind us to the innovation that we can still have and we can see in a more diverse, competitive marketplace.

So I think we need to take the learning from what we've seen, but we also have to look forwards.

Lord Tim Clement-Jones: I agree with Chris. I mean, the European Union, yes, it has a set of big tech companies, SAP and so on, but it doesn't have anything in the internet or AI space. So, the concentration is increasing. It isn't just concentration in the United States. It's concentration with US companies throughout the world.

And that's what people are finding really difficult because they are getting squeezed out of their own markets, and it's making market entry that much more difficult for their own competitors.

So, although yes, we've seen innovation, the question is, is that innovation gonna stifle competition if we're not careful? And will we end up with a situation where we only have a handful of US based companies that are making all the running?

And that is not a situation which I think is gonna be healthy in the end, both economically or politically, or in terms of people feeling confident that they understand what they're getting from the tools that are being provided to them.

Michael Krigsman: All right, we have another really interesting question. This is from Shelly Lucas on LinkedIn. And Shelly says, "If we are entering potentially an era of circular AI generated content, how do you ensure AI alignment, especially considering the pace at which models are being updated with diverse and sometimes conflicting safety standards?"

Lord Tim Clement-Jones: Well, I mean, this is the extraordinary thing. We don't know where we're gonna end up with this, partly because it's still early days with quite a lot of the large language models, but they've been trained on the internet, and so we don't know actually how much garbage in is gonna mean garbage out, in a sense.

We don't know what the hallucinations will look like over time, whether they'll increase. I've seen papers which suggest they may, and what the circularity is going to do, when AI is actually, if you like, you feed it a lot of AI generated material, what does that do to the underlying algorithm?

So yes, this is a really, really complex issue, and of course, the challenge is that you don't really understand what the companies, the tech companies are doing because they're extremely opaque in how they actually go about their business. And it's very difficult to get transparency.

So where does the transparency come in, and how do you monitor what's happening with the large language models?

Lord Chris Holmes: This is where transparency comes in, accountability. This is where you get this access. It's why there is no alternative to the regulator being funded, having the skills and the know-how, and getting in there and doing the work.

But Michael, this is where I really believe we can't just leave this to the regulators alone because it isn't as if there's enough of them or they're well funded enough or they've got enough skills and know-how. It's where we come back to this absolute necessity, imperative of every business, every individual upskilling, getting knowledgeable.

That's the only way that we are gonna have a chance because there aren't enough regulators and there won't be enough regulators, unless they've got kind of multiple billions of them. So I think we've got to really take this on as an individual, business, community, society, and country-wide responsibility.

Yes, we need the regulators, but as individuals and businesses, we absolutely can't leave it to that alone.

Michael Krigsman: But Chris, I have to say, you know, you talk about upskilling. We have a situation where the average business leader, they're not technologists. They don't know anything about artificial intelligence. The regulators don't know anything about artificial intelligence.

And we've got these leading companies that are so incredibly powerful and competent and they have brilliant marketing, and the average person who's trying to adopt AI, they know nothing.

Lord Chris Holmes: And therein, Michael, is where the problem lies and therein is where the answer sits. Because there are brilliant explainers out there. We can get everybody, you can do an hour's worth of upskilling in AI and probably be about 60, 70% there.

So, upskilling's possible. Getting knowledgeable is entirely possible. Yes, you may not be able to break into the black box, but you can get that level of information and knowledge. So, I think that is entirely possible. And then regulators, they can be upskilled too.

So we just absolutely have to do that work. We've got to put that effort in. I believe it's possible. I'm convinced we can do it. We just have to go and do it. That's the piece.

Michael Krigsman: I also want to go back to Arsalan Khan's question. He also followed up and he said, "Given the concentration, maybe US tech dominance is actually less risky than other countries getting it." But I think that's an interesting perspective that I want you both to address.

Lord Tim Clement-Jones: Well, I mean, it's all a question of degree. What I would say is, I mean, for instance, the UK is not gonna have a frontier model itself. There's no frontier model being developed in the UK. There are foundation models, open source. And there are lots of applications, but there's no frontier model.

So the question is, for the UK and for other countries, what does access look like? You know, what are the quid pro quos? Is it, for instance, big US tech company comes in and buys a huge data center, does that give it a kind of permission, if you like, to operate without restriction, regulation, and so on? And so we have to look at this very carefully.

I mean, my view is that if you look at all the US tech companies, they're not the same. I mean, Meta with its open source LLaMA is very different from a number of the closed source frontier models. There are ethical differences between them.

And so, I don't think you should lump them all together, but you've got to look at each of them carefully and say, what is their ethical stance? What do they stand for? You know, are they transparent? Are they accountable? Those kinds of issues. And then you calibrate your response accordingly.

But the one thing we can't do is let it rip and not have any kind of regulatory structures at all, which is where I fear the new administration in the United States is gonna go, unfortunately.

Lord Chris Holmes: This is where I think we absolutely have to understand, and this is a global understanding that we have to have, that if we don't have good regulation, actually, we will stifle innovation. We'll stifle economic growth. And we certainly won't empower our citizens.

So when we get good, effective, proportionate, trusted, transparent regulation, people get on board, people adopt the technologies. And then that drives your economic growth, that drives innovation.

But if we don't have that, then it atrophies. We don't have the citizens engaged. We've lost the battle already. So, this is about us understanding that regulation isn't some external thing that sits to one side. It isn't a battle between regulation and innovation and growth.

Actually, the right kind of regulation, properly done, enables trust. Trust enables adoption, which equals economic growth, which drives everything forward. So we have to understand those dynamics. The regulation doesn't have to be the anathema, the thing to be slain and got rid of. Good regulation drives growth.

Michael Krigsman: I keep coming back to how you enforce these regulations because I'm really curious about that. Because now we have this situation where you have, again, the incredibly powerful companies who know what they're doing, and you have people who are trying to learn. And you have both made the point that we need to educate ourselves, and that's absolutely true. But the reality is, the vast majority of people are simply not gonna do that.

Lord Chris Holmes: And I'm not sure we should fall on that. I'm not sure we should succumb to that reality, though, Michael, because, look, we managed to get pretty much everybody vaccinated during COVID. So we can do these things. We can. We achieved that. We can do this. We can upskill. We can get people engaged. It absolutely is possible.

And I actually think for the health of our society, for the health of our democracy, for the health of our economy, we have to do this. We can't accept that, "Oh well, nobody's gonna get upskilled, nobody's gonna get engaged." This is too important. So we have to do this. And I genuinely believe we can. We just have to apply ourselves to it.

Lord Tim Clement-Jones: I mean, I think Chris is really, I mean, he's right and he's an optimist, which is great. I mean, we need to get upskilled. But the fact is that most people are not gonna have unlimited time to go away and study all this stuff.

So, what we need is trusted sources of information. We need proper public broadcasting. We need proper independent journalism. We need sources that give people accessible information that don't give them the Big Tech line, necessarily.

And I think that we can't rely solely on individual upskilling, although yes, in a sense, a little knowledge can go a long, long way, and Chris is absolutely right about that. But it's trusted sources, and that's one of the reasons why the impact on, if you like, traditional media is so important.

Because if that disappears, then we have a real problem because we no longer have those independent sources. And that is one of my big worries in all of this.

And of course, there's an issue beyond these kind of major models, which is in a sense the weaponization of AI for bad purposes. And that's why I chair a group called Peers for the Planet, which we're looking at things like deep fakes and misinformation. That is another hugely important area where we need regulation.

Michael Krigsman: All right, we have another question from LinkedIn. This is from James Merriweather. And James asks, "How do these issues affect AI governance and risk management?"

Lord Chris Holmes: I think it's got everything to do with it because if we don't have the understanding at the board level, at the strategic level within an organization, around what is the plan, what are the opportunities, what are the risks, then an organization won't be able to effectively manage that risk.

So, it is about this entire upskilling all the way through an organization, right from the board downwards. You know, what are the opportunities? What are the risks? And how do you effectively manage that?

And that's why this isn't something that just one person in an organization should be doing or should be responsible for. This has to be something where everybody understands what is the plan, where are we going, and what are the potential pitfalls.

So, I think it's got everything to do with governance and risk, and you've got to manage it accordingly. And if you don't have the understanding and you're not upskilled, you can't manage that effectively.

Lord Tim Clement-Jones: I'd just add to that because the question actually is more than just the individual business. It's also the public sector. You know, how is the public sector addressing this? What kind of boards, if you like, have they got overseeing AI? What are the governance structures?

And I think this is a massive issue because so much is being relied upon AI now to deliver public services. So, yeah, I think governance is, if you like, it's right across the board. It's not just private sector. It's also our public services and so on.

And I think we need to be extremely careful how we're deploying AI and making sure that there are proper governance structures in place for that. And you know, we saw what happened with the post office Horizon scandal in the UK, where basically a computer was believed over human beings, and it ended up with huge miscarriages of justice and real tragedy for people and their lives.

So, we need to be very, very careful that we get the governance right with AI as well.

Michael Krigsman: We have a question from Twitter, and this is an interesting question. I'll address this to you, Chris. But John Thornburgh says, "Is there any recourse if or when, really, when LLMs start citing each other?"

Lord Chris Holmes: So, look, John, this is where the transparency comes in. This is where the accountability. This is where you mandate these things. So that the tech companies have to be open on what is their training data. Where has it come from? And they have to open on what sources. Where are they getting the material?

And then I think you've got to have the ongoing monitoring and oversight of that. Because if you leave that to its own devices and you've got no insight into it, no transparency, no accountability, then yes, it's the circularity, as we've already touched on earlier, and we don't know where that goes.

And that is potentially entirely dangerous, entirely dystopian, and deeply worrying. So this has got to be mandated that you have to have that transparency and you need to have the accountability.

Lord Tim Clement-Jones: Yeah, and I mean, I would just add to that. I mean, this is a question really of regulation. You know, what are the regulations which mandate transparency? I mean, the AI Act has got quite a lot in terms of transparency requirements around high risk AI.

The UK is not gonna have that same kind of structure, but I think proper AI regulations should have those transparency requirements built in. And one of the issues at the moment is we allow too much secrecy. And it's almost because these companies, they claim all their trade secrets, commercial confidentiality, whatever. They don't have to reveal things.

And actually, we need much more openness in terms of where's your data coming from, which Chris just said. What's the provenance of all of that? Where is it coming from? And we should be insisting on that.

Michael Krigsman: You know, just following up on this issue of regulation and enforcement, I mean, in the United States, the argument is being made by the current administration that we want to ensure that we can compete with China, and therefore, we should limit regulation as much as possible. So how do you balance, even thinking about it from a US perspective, how do you balance those issues of competition with the regulatory issues that you're both raising?

Lord Tim Clement-Jones: I mean, I just think that's a false narrative. It really is a false narrative because the fact is, I mean, if you have no regulation and you have really irresponsible use of AI, then actually you're gonna end up creating distrust. You're gonna end up with people not wanting to use the products. You're gonna end up not creating trust and confidence in your own population or other populations around the world.

So, I think that we have to take a much more responsible line on this. And actually, the competition is not about the absence of regulation. The competition is about responsible AI, which actually is trusted and which people want to use and where there's confidence. So, I think that's really the way of looking at it.

Lord Chris Holmes: I couldn't agree more, Tim. I mean, this is a fundamentally flawed narrative, that in some way, there's a choice between innovation and regulation. And it goes back to what I was saying earlier, that actually, if we get regulation right, in a proportionate way, which is trusted, transparent, effective, it drives trust. Trust drives adoption. Adoption drives innovation, drives economic growth.

So we have to understand these dynamics. They aren't separate. They aren't in opposition. They're entirely interrelated. And if we get the regulation done in the right way, if we do it for the right reasons, this drives growth. This drives what we want.

So, this is not a choice between innovation and regulation. And I think if people are trying to position it in those terms, then they're either not looking at the evidence, or they've got some other angle.

And I think businesses, people, countries, we need to step out of this entirely false dichotomy because that isn't the reality. The reality is, you get good regulation, you drive growth, you drive innovation.

Michael Krigsman: So, with that, let's take a short break, and when we come back, we're gonna talk about some of the specific issues around regulation and how they relate to business leaders. So, everybody, hang tight, and we will be right back.

Copyright, Open Source, and Data Rights

And we're back with Lord Tim Clement-Jones and Lord Chris Holmes. We're talking about the AI oligopoly, Big Tech, regulation, enforcement. And I want to talk more deeply about a couple of specific issues.

And there are two that I'd like to focus on. One is the issue of copyright and the scraping of data, how AI companies are using data. And the other issue relates to open source and the role of open source in this broader AI ecosystem. And so let's start with copyright. Tim, you care about this issue a lot. Why?

Lord Tim Clement-Jones: Because I'm a lawyer, first of all. I used to be an intellectual property lawyer, and this is at the heart of my life, in a sense. And you know, the question is, are creators gonna get rewarded in the age of AI?

And what we've seen is that a lot of the models have been trained on whole swathes of the internet without permission, without payment. And you know, the tech companies are asserting that this is fair use or fair dealing in our jurisdiction, and that creators should suck it up.

But actually, that means that there's a massive misalignment between the value that's being created by the AI companies and the fact that the creators themselves, whether they're journalists, photographers, artists, musicians, they're not being remunerated for their content being scraped. And I think that's fundamentally wrong.

And I think we need to develop a system whereby creators are adequately remunerated. Now, there are various models out there. But the principle is that you shouldn't be able to freely scrape without permission and payment. And that's the key issue.

And I think in the UK, we're looking at a code of practice, which actually doesn't have that, doesn't go far enough, unfortunately. And so we're pushing the government to do more on this. And I think it's gonna be a huge issue in the years ahead because creators are very angry about what's been happening.

Lord Chris Holmes: And this is where you absolutely need fairness. You need equity built into this. And this comes back to the regulation, comes back to accountability, transparency. And I think you're absolutely right, Tim.

When you look at people's livelihoods, people's life's work, their creativity, their artistry, all of that is part of their identity, part of who they are. If you effectively allow that to be taken without permission, without payment, without recognition, then that isn't fair. That isn't equitable.

And I think what it does is it tells you something about the nature of the approach by some of these tech companies. And that's one of the problems. So I think we need to see a fundamental shift in this.

We need to see that creators not just have some notional recognition. They're actually remunerated appropriately. They're acknowledged, and their work is dealt with appropriately. This can't be something where we just have this wild west land grab approach. That isn't acceptable.

And if we allow that to continue, then it's just further evidence of the winner takes all approach, which goes back to what we were talking about at the very outset of our conversation, Michael.

Michael Krigsman: So, speaking of land grab, what about open source? And Meta has made a big deal out of, we're open sourcing. We're sharing. But at the same time, I think a lot of people question the motivations. What are your thoughts on the role of open source, and to what extent should that be encouraged?

Lord Tim Clement-Jones: Well, it's interesting because Meta will say, "We're actually being beneficent, we're offering this to the world." But of course, Meta doesn't have at the moment a world-beating closed source model, a frontier model. So in a sense, it's in their interest to undercut through open source the closed source organizations. So it's not entirely philanthropic in that sense.

But I think open source has got huge benefits. I think in a sense what we want is a level playing field. We want both proprietary models and open source models to be regulated in the same way. And I don't think we should give open source a free pass.

I think we need to make sure that there is a regulatory structure around open source as well. And there are various approaches to that. But I think it is, it's not a question of whether you should be in favor of open source or not. Open source has been hugely beneficial to the world over the years. But equally, we need to make sure that it's properly regulated.

Lord Chris Holmes: And Michael, I'd say a couple of things on this. So, I mean, open source is clearly important. It does have a role to play. But let's not have any naivety about some of the motivations from some of the companies. As Tim said, it isn't entirely clear or clean in how and why they're doing this.

But I think that openness, that access to technology, that ability for people to experiment, that ability for the smaller players, the innovators, to be able to come in, open source has a real role in that.

But I think we've got to do it with eyes wide open and we've got to understand what are the potential risks and issues here. And you go back to the transparency and the accountability once again. If you've got open source models out there, it doesn't mean that therefore anything goes. You've still got to have some controls, some guard rails in place.

So, I think open source, yes, but it's not without its complexities and issues that we need to think carefully about.

Michael Krigsman: There was a report recently, it was from the Stanford Human-Centered AI group. And they did a foundation models transparency index. And what they found is that essentially all of the companies scored very poorly in terms of transparency, but the open source companies actually scored even worse than the closed source companies. What does that suggest to you?

Lord Tim Clement-Jones: Well, it just shows that the fact of putting your code and so on out there doesn't necessarily mean that you've got that transparency in terms of the provenance of your data or even the safety standards that you've applied to the development of that foundation model.

So I think it just illustrates that open source is not a panacea. You actually do need those transparency measures right across the board. And I think the UK government is looking to do more on that. But I think open source shouldn't get away with less transparency than the closed source models.

Lord Chris Holmes: And I think that's absolutely right. And I think what it tells you is that open source isn't the answer to all issues. Open source still has to be subject to regulation. There still have to be transparency measures.

And I think that tells you something about what is required across the board, whether you're open or closed source. This doesn't give you a free pass. This doesn't mean that you can just do anything.

So I think that's a really interesting finding, and it's something that should definitely give pause for thought for governments and regulators as they're thinking through, how do we deal with all of this? And what are the different requirements for the different types of models and companies?

UK's AI Strategy and Economic Concerns

Michael Krigsman: So, let's shift gears and talk about the UK specifically. The UK has ambitions to be an AI superpower, a leader in AI. And you're both lords in the House of Lords and involved in policy. What do you think about the current state of the UK's AI ambitions and where the UK is?

Lord Chris Holmes: So, I think there's a lot to be positive about. There's a lot of great work going on. There's a lot of great technology, great science going on here in the UK. We've got to make sure we support it. We've got to make sure we create the environment. Regulation is a part of that.

But we've also got to make sure that our researchers, our developers, the innovators here have access to compute. They have access to data. They have access to the skills. And we've got to create that complete ecosystem that enables them to flourish.

So, I think there's a lot of potential. But potential, as I said right at the outset, isn't inevitability. So we've got to do the work to support it. And I think it's really incumbent on government to work with business, to work with research, to work with the tech sector, and all of us together to create that environment where we can see all of those good things flourish. But it isn't a given. We've got to do the work.

Lord Tim Clement-Jones: I entirely agree with that. And I think the UK has got huge strengths in AI, and not just frontier models. We've got huge strengths in AI applications right the way across the board. But I think we've got to be careful. We've got to be extremely careful about a couple of things.

One is how much access is the UK actually gonna have to some of these major US models and will we have sovereignty over that? And secondly, are we just gonna have our lunch eaten in terms of US companies coming in and basically squeezing out the UK companies? And I think we've gotta be very careful about that.

I mean, one of the big issues is data centers and compute. And the UK government has a plan around that, but we've gotta make sure that we actually level the playing field and we don't just give the Big US tech companies a completely free pass. So, I think we need to make sure that we support our homegrown tech as well as being open to the US companies coming in.

International Competition and Geopolitics

Michael Krigsman: Let's talk about China for a moment. How do China's AI capabilities compare, and to what extent should that be a concern? Is that a reason to limit regulations as the US administration is suggesting?

Lord Tim Clement-Jones: Well, I mean, that's, I mean, China is extremely good at AI. And they've got their own models. DeepSeek, for instance, is an extremely good model. And they've put huge investment in. They're putting huge amounts of money into AI chips and so on. And they're not sitting still. So I think we have to recognize that.

But the question is, we have to make a decision about what kind of world we want. And China has got a very different value system, a very different governance system. And I think we have to be very careful that we don't kind of blindly adopt, if you like, a China model or an approach which doesn't reflect our own democratic values.

So I think we've got to be, we've got to be pretty careful. We've got to compete. We've got to be good at this. But equally, we've got to make sure that we've got the values embedded in the way that we develop and deploy AI. And so it isn't a free-for-all. It's actually, it has to be done in a responsible way, which reflects Western democratic, if you like, norms.

Lord Chris Holmes: And I think, Michael, this is where it's really important to understand that there is no race to AGI that is outside of the context of society. There isn't this binary kind of race to AGI. We win, and China doesn't. Or China wins, and we don't.

We have to think about this in the context of what type of society do we want? What are our values? What are our democratic structures? And how does this embed within that?

So I think it's really important that we don't position this as, oh, there's this arms race, this AI race. We have to drop all regulation. We have to just let it go completely and then we'll win. And then we'll deal with the downsides later. That doesn't work. That doesn't work with any technology. That doesn't work in the real world. That doesn't work in our societies.

So we have to think about how does this technology and these innovations fit within the society that we want, the values that we have, the democratic structures that we have? And if we do that, then we will succeed.

But if we just go down this route of, it's a race and all regulation is bad and let's just go hell for leather, then we will lose. We will lose our values. We'll lose our democratic structures. And actually, we won't win.

So I think we have to think much more carefully about what does winning actually look like and what do we want our world to look like.

Michael Krigsman: We have a question from LinkedIn. This is from Wayne Anderson. And Wayne asks, "Should there be international bodies or agreements to regulate AI globally, similar to other international frameworks?"

Lord Tim Clement-Jones: Well, I think that would be, in an ideal world, fantastic. The problem is, of course, at the moment, we've got a very fractured geopolitical landscape. And I'm not sure whether that's actually realistic in the short term.

I mean, the UK held a couple of AI Safety Summits, Bletchley Park and then in Seoul, trying to get some international cooperation. And I think there are various initiatives, but I'm not sure they're hugely effective at the moment.

And I think what we're gonna see is regional and national approaches. And I think the challenge will be, can we, in a sense, create enough commonality between, say, UK, EU, US approaches that we've got some kind of harmonization so that businesses can actually operate across these jurisdictions without having completely different regulatory landscapes?

So I think that's probably the best we can hope for in the near term, rather than some grand international convention.

Lord Chris Holmes: And I think, Wayne, it's a great question. And I think Tim's right. I think we have to be realistic about what's achievable in the near term. But that doesn't mean we shouldn't be striving for it in the medium to longer term. I think we absolutely should be.

But in the here and now, I think what we need to do is make sure that there's as much commonality, as much learning, as much sharing of best practice as we possibly can. And I think it's really important that regulators, that policymakers are talking to one another, are sharing what's working, what isn't working, and really trying to build that common ground so that we can over time, potentially get to some kind of international framework or agreement.

But I think we have to be realistic that in the short term, it's probably going to be more regional, more national. But that doesn't mean we can't be working towards something more harmonized over time.

Compute, Infrastructure, and Market Dynamics

Michael Krigsman: Let's talk about compute and infrastructure for a moment. And in the United States, Elon Musk has essentially said the US and China are in a race to build these massive data centers to train AI. And whoever gets there, wins. Your thoughts on that?

Lord Chris Holmes: Well, I mean, I think we have to see this in terms of the infrastructure that's required. Compute is clearly hugely important, access to compute, access to data, access to energy. All of these are part of the infrastructure that's needed.

But I don't think it's quite as simple as just whoever builds the most data centers wins. I think it's about how do you deploy those resources? How do you use that compute? What are you actually trying to achieve with it? And does it align with your values, with your societal goals, with your democratic structures?

So, yes, we need the infrastructure. Yes, we need the compute. But we also need to be thinking very carefully about how we use it and for what purposes and what are we trying to achieve. So I don't think it's quite as simple as just a race to build data centers. It's much more nuanced than that.

Lord Tim Clement-Jones: I agree. And I think the energy consumption involved in this is massive. And we've got to be really careful, environmentally, about what's happening here. I mean, there are huge, huge energy demands, and we're seeing investment in all sorts of energy sources, including nuclear, to try and power these data centers.

But that's gonna have environmental impacts. And we've got to make sure that we're factoring that into our considerations about what's the right level of investment in compute and data centers. So, yes, it's important, but it's not the only consideration. And we've got to look at the broader picture.

Michael Krigsman: There's another question that came in from LinkedIn. James Merriweather asks, "Is compute limiting innovation, or is it more about access to training data?"

Lord Chris Holmes: I think it's a bit of both, James. I think clearly compute is a constraint for many, particularly smaller players, smaller innovators. They may not have access to the compute that they need. But equally, data, quality data, the right data, that's also a huge issue.

So I think it's probably a combination of the two. And I think what we need to do is make sure that we're creating an environment where there is access to compute, where there's access to good quality data. And I think there are various initiatives around shared compute resources, public data sets and so on.

But I think we've got to make sure that we're enabling the smaller players, the innovators, to have access to both of those things. Otherwise, it just further entrenches the position of the big players. So I think it's both, and we need to address both.

Lord Tim Clement-Jones: Yeah, I mean, I think Chris is right. And one of the issues is gonna be synthetic data, how much can we use synthetic data to train models? That's gonna be interesting. And the other thing is, what happens with the kind of trade in, if you like, data between different organizations? Can we create data marketplaces and so on?

I think there's a lot of innovation that could happen in that space. But it needs some regulatory certainty, actually, to allow that to happen. So I think data is really critical. Compute is really critical. But we've got to create the right environment for innovation around both of those.

Investment Concentration and Economic Risk

Michael Krigsman: We have time for maybe a couple more questions. There's a question from LinkedIn. Todd Lohenry asks, "In the frothy AI economy with massive investments, when does it become a bubble? And are you concerned about that?"

Lord Tim Clement-Jones: Well, I am concerned. I mean, I think the level of investment is huge. The valuations are huge. And we don't know yet whether the business models, particularly some of the big AI companies, whether the business models are sustainable. Are they actually gonna make money? Are they gonna be able to sustain these valuations?

And if not, then we could be looking at a pretty major bubble. And I think we have to be prepared for that possibility. And governments need to be thinking about what happens if there is a major correction in this market? What are the implications for the economy more broadly? So, yes, I think it is a concern.

Lord Chris Holmes: It's something that a lot of us should be at least somewhat concerned about that economic concentration and the circularity of funding, and applying all the principles that we know. So looking as to what you're investing in, what is the return on investment? Applying all of those good economic and business principles that are well-made will give us the right answers and the right analysis. And of course, this is gonna be one great bubble at the end of the day. I think we're all a little bit apprehensive about this.

Michael Krigsman: Chris brought up this issue which we have not touched on, which is the circularity of funding. Just your views on that.

Economic Interdependence in AI Development

It's such an important issue and we, but we're out of time. Yes, absolutely.

Lord Tim Clement-Jones: And this is exactly why people are worried about this bubble because Microsoft is investing in OpenAI. OpenAI is buying credits from Microsoft. And Nvidia is doing pretty much the same. And so they're feeding each other economically.

And if confidence goes and if OpenAI doesn't make any money and isn't making a great deal of money at the moment, then what happens? Everything goes bust and it all depends on Microsoft continuing with its mainstream business rather than with the large language model ChatGPT.

Advice for Business Leaders in the AI Era

Michael Krigsman: What should business leaders and boards of directors do in this complex environment with shifting regulations, changing technologies? Your advice, please, for business leaders and boards?

Lord Chris Holmes: Get involved, get upskilled, get knowledgeable. Get involved in a debate, a discussion. Understand what you're trying to achieve as a business. Understand how, potentially, these technologies could help and partner up and really connect and make it a team effort.

Lord Tim Clement-Jones: Know what your values are in the middle of all of this because there's nothing like AI for making you question your values in all of this.

Michael Krigsman: Know what your values are. That's an abstract concept potentially.

Lord Tim Clement-Jones: It is potentially abstract. But, frankly, what is the purpose of your business? How do you treat your employees? What kind of trust do your consumers, do your employees and your other stakeholders have in you? Well, often that's all about the values that you as a business have. And I think we forget about them too often.

Guidance for Regulators and Policymakers on AI

Michael Krigsman: Chris, what advice do you have for regulators and policymakers when it comes to this set of AI issues?

Lord Chris Holmes: Understand what your role is. Understand how these technologies are developing and how they're already being deployed. And look about an approach where you do right size regulation. And you understand the complexity of the needs of the consumer, the creative, the citizen, the investor, the innovator.

It's not straightforward, but it's completely doable. And we've seen examples in other areas in the UK where we've done it. So we know how to do this. We need to do it. And in the words of a great American company, their best ad campaign ever, Nike, ultimately having done all that, "Just do it."

Michael Krigsman: Tim, you're gonna get the last word. Advice for regulators and policymakers.

Lord Tim Clement-Jones: Resist the blandishments of big tech who will tell you that all regulation is terrible. They won't be able to innovate and do business. But you have to resist that. We know the resources that are going into this from big tech both in Brussels, London, Washington and so on. And we need to be resistant to that.

Michael Krigsman: How do you resist? They have so much money and resources to throw around and share. How do you resist?

Lord Tim Clement-Jones: You resist by coming on CXOTalk and influencing the influencers, Michael.

Michael Krigsman: That works for me. And with that, a huge thank you to Lord Tim Clement-Jones and to Lord Chris Holmes. I am so grateful to you both. Thank you for taking your time and sharing your expertise with us once again.

Lord Tim Clement-Jones: Thank you very much.

Lord Chris Holmes: Pleasure as always.

Michael Krigsman: And I hope you'll both come back. Everybody, thank you for watching. Now before you go, subscribe to the CXOTalk newsletter. Go to cxotalk.com. We have amazing shows coming up. Check it out. We'll see you again next time, everybody. Have a great day. And have fun with AI.