India and UK stood up to Grok against generation of obscene content: UK minister - The Economic Times

India and the UK took a stand against the generation of obscene content by Grok, UK minister for AI and online safety Kanishka Narayan told

Surabhi Agarwal in an interview, underlining that the power of the nations to “ground AI development” in shared values “is not to be underestimated.” Narayan, who is of Indian origin, said he was impressed by the country's small, efficient language models adapted to the vernacular experience. Edited excerpts:

The first AI summit took place in the UK. Since then, the world has changed quite a bit. Is AI a threat or an opportunity from your lens?

Yes, we started the journey at Bletchley Park with a range of countries, and India was a keen partner for the UK throughout. We're absolutely excited about the focus of this summit on “impact.” That impact connotation really captures the way I think about this, which is (that) it's not either opportunity or risk, but actually a nuanced sense of pushing very fast and hard on where the opportunity is, spread prosperity across the country, make sure people feel dignity in AI usage. But alongside that, looking at risks as well and making sure that we're being responsible in the adoption of AI.

Your Prime Minister was here recently and India and the UK announced a big partnership. What can we expect going forward in the area of AI?

That's a really important point because the UK and India were amongst the first to get to an FTA that was so impactful economically for us. But what was very unique about it was that innovation was a distinct chapter in it. It was very clear, recognized on our side and the Indian side, that innovation and, at the heart of it, technology and AI, were central drivers of this relationship. We have furthered that in the technology security initiative, and I think that forms a really fantastic foundation for it. In terms of specific areas of focus, from my point of view, I think how we can collaborate on aspects of infrastructure, including chips. ARM, one of our leading chip design companies, already has a large presence in Bengaluru, great talent working to make sure that great job and a sense of expertise developing here in India alongside the expertise that we've had in the UK context. That's an important area of focus.

(On) the question of models, one of the observations I've had just in the last 24 hours here in New Delhi, is that India is doing a really interesting job at small, efficient language models adapted to the vernacular experience of people. That speaks to our focus on adoption as well and so there's huge amounts we can learn and engage with on that. On the other hand, as I mentioned, we have a focus on science. India has a huge population that is focused on making sure that it is healthy, it is happy. So our advances in drug discovery are really important. And all of this is really grounded in a long-lasting relationship that we've had. If you look at, for example, vaccines. There's great groundbreaking research at Oxford in the context of Covid vaccines, with great contributions from Indian companies, and then manufacturing in Wales as well. And so we want to do exactly that. Bring the strengths of the UK and India together and really push the frontiers of AI.

You grew up in India. You understand the country well. What would be your advice to counterparts in India when it comes to the positioning of AI because, increasingly, AI is being used as a geopolitical tool by nations...

The first thing to say is, I will speak from a position of humility. I did spend a lot of time in India, but India has changed at a rapid pace and keeps changing. So I'm in no position to give advice. But here's some words of experience from our point of view. You have to focus fundamentally on your shared values. We saw this at the start of the year, where Grok was allowing the nudification of women's and children's images. India stood up and said this is not right by our values, we will not accept it. The United Kingdom stood up and called Elon Musk out and said this is not in line with our values, (and that) we need to stop it. As a combination of that, it stopped. So the power we have together to ground our AI development in the values that we share, and as a result, move the world, is not to be underestimated.

The UK has taken a different approach when it comes to AI regulation from the EU. What are your thoughts?

The two things I say on the regulatory front… One, the quality of regulation is only as good as the quality of the people in the regulatory context and in the government context. And so we have been relentlessly focused on the quality of people we have thinking about this question. The AI Security Institute in the UK is unparalleled in the quality of its people and the extent of its access to models to be able to understand where the risks are. And that's a really fundamental starting point--good capability in the public sector beats the quality of regulation every single time.

The second aspect of it is we have said where there is, in a sector specific way, a risk, or where there is an acute risk, let's say child abuse material, for example, we are not going to compromise on those questions. And so we have gone very robustly in both sector context and through aspects like the Online Safety Act, to say where the risks are here, we will regulate them. Where the risks are hypothetical or potential or prospective, we will build capability, we will dynamically move, and, yes, if needed, we will move in the direction of stopping those risks.

The UK is positioning itself to be an AI superpower. From that standpoint, what is your strategy?

We had an AI opportunities action plan over a year ago, and we announced that even just within a year, we've achieved 75% of it. I want to try and close the final 25% very soon. And so in terms of the UK's positioning as an AI superpower, I think there are three things that are really top of mind. One, being right at the frontier. Google DeepMind started in London, so a huge number of the advances remain focused at the frontier, and we're building great models in science and drug discovery and materials discovery, making human progress grounded in the UK.

The second part is we want to be the best country in the world at adoption. We have a particular focus on what creates value for people in the country. And the thing I'm passionate about in the Indian context is the Prime Minister's focus on welfare for all, happiness for all. We share that deeply in the UK context, and we want to do that through the medium of adoption of AI.

Then the third way in which we're positioning globally is being very thoughtful about responsibility in the adoption of AI as well. We have a great capability in government, the AI Security Institute looks at that. But more broadly, we think we can get the social policy, the institutions right to be able to make sure AI benefits all of us.

There's concern about the concentration of power in AI in the US and China. Does that worry you? That some countries are getting left behind? How do you think they can catch up?

One thing that I think all of us are recognizing is that AI is increasingly the frontier for both economic power and hard material power as well technologically. What we're focused on is making sure we're deepening our alliances. The US has been a great partner for us. We have not just an economic agreement more generally, but a very specific technology prosperity deal as well, which has deepened the US and UK relationship. We want to keep doing more of that with India as well. And I think in that context, the UK is in a very distinct position because, by virtue of where we've arrived, which is a position of deep trust with the US, with India, with allies, I think we're now increasingly a safe harbour.

If you're a US company in a sensitive area developing technology, the UK is a totally trusted place now. You should come and scale in the UK, and many of them (these companies) have. If you're an Indian company and you're thinking, ‘How do I partner with others in a country that is trustworthy, that we share values with?’ The United Kingdom should be top of your list. In this evolving geopolitical moment, the UK's history of relationships and pragmatism in deepening those will serve us very well.

What is your view on sovereign AI? More countries are pouring billions into having GPUs within the country. What is the UK's approach to that?

Sovereignty is a very ambiguous term that people use. There are three things that I think are critical to it. One, have enough critical inputs for AI. So yes, we want to get GPUs in the UK, and if that means partnering with companies we trust, we will get that. The second is to diversify your engagement with companies as well. One of the things I'm excited about here in the Indian context, is working with companies in India, fantastic innovators coming to the UK as their first international expansion. And then the third thing that is really critical is building domestic capabilities. We have amazing chip design startups doing AI for inference. We have great science model builders doing real frontier work. And we have some great people doing applied AI work as well. We've got a great delegation that I encourage everyone to look at. We've got great companies here as well. Each of those steps to sovereignty have enough of the inputs, diversify those inputs, and then, yes, build domestic capability. We're walking up each of those rungs of the ladder.

Tell us a little bit more about the areas where you think that India can contribute in the area of AI and where the UK can contribute toward building a greater AI ecosystem in India?

India has an exceptional tech ecosystem from a talent point of view. That talent building models can be quickly and economically efficiently adopted. I've just been speaking to some of your founders. They're building models that are small models that can be deployed in the palms of people's hands rather than just large data centre clusters. In the opposite direction, we have a great university system. Four of the world's top 10 universities (are) in the UK, a very R&D-intensive AI ecosystem. We want to make sure that we're using that to spread opportunity for people right across India as well. There's some fantastic British universities out here already making sure that brilliant, energetic Indian talent gets the benefit of that support as well. So I think those are two specific areas.

Dia Rekhi contributed to this story.

Surabhi Agarwal in an interview, underlining that the power of the nations to “ground AI development” in shared values “is not to be underestimated.” Narayan, who is of Indian origin, said he was impressed by the country's small, efficient language models adapted to the vernacular experience. Edited excerpts:

The first AI summit took place in the UK. Since then, the world has changed quite a bit. Is AI a threat or an opportunity from your lens?

Yes, we started the journey at Bletchley Park with a range of countries, and India was a keen partner for the UK throughout. We're absolutely excited about the focus of this summit on “impact.” That impact connotation really captures the way I think about this, which is (that) it's not either opportunity or risk, but actually a nuanced sense of pushing very fast and hard on where the opportunity is, spread prosperity across the country, make sure people feel dignity in AI usage. But alongside that, looking at risks as well and making sure that we're being responsible in the adoption of AI.

Your Prime Minister was here recently and India and the UK announced a big partnership. What can we expect going forward in the area of AI?

That's a really important point because the UK and India were amongst the first to get to an FTA that was so impactful economically for us. But what was very unique about it was that innovation was a distinct chapter in it. It was very clear, recognized on our side and the Indian side, that innovation and, at the heart of it, technology and AI, were central drivers of this relationship. We have furthered that in the technology security initiative, and I think that forms a really fantastic foundation for it. In terms of specific areas of focus, from my point of view, I think how we can collaborate on aspects of infrastructure, including chips. ARM, one of our leading chip design companies, already has a large presence in Bengaluru, great talent working to make sure that great job and a sense of expertise developing here in India alongside the expertise that we've had in the UK context. That's an important area of focus.

(On) the question of models, one of the observations I've had just in the last 24 hours here in New Delhi, is that India is doing a really interesting job at small, efficient language models adapted to the vernacular experience of people. That speaks to our focus on adoption as well and so there's huge amounts we can learn and engage with on that. On the other hand, as I mentioned, we have a focus on science. India has a huge population that is focused on making sure that it is healthy, it is happy. So our advances in drug discovery are really important. And all of this is really grounded in a long-lasting relationship that we've had. If you look at, for example, vaccines. There's great groundbreaking research at Oxford in the context of Covid vaccines, with great contributions from Indian companies, and then manufacturing in Wales as well. And so we want to do exactly that. Bring the strengths of the UK and India together and really push the frontiers of AI.

You grew up in India. You understand the country well. What would be your advice to counterparts in India when it comes to the positioning of AI because, increasingly, AI is being used as a geopolitical tool by nations...

The first thing to say is, I will speak from a position of humility. I did spend a lot of time in India, but India has changed at a rapid pace and keeps changing. So I'm in no position to give advice. But here's some words of experience from our point of view. You have to focus fundamentally on your shared values. We saw this at the start of the year, where Grok was allowing the nudification of women's and children's images. India stood up and said this is not right by our values, we will not accept it. The United Kingdom stood up and called Elon Musk out and said this is not in line with our values, (and that) we need to stop it. As a combination of that, it stopped. So the power we have together to ground our AI development in the values that we share, and as a result, move the world, is not to be underestimated.

The UK has taken a different approach when it comes to AI regulation from the EU. What are your thoughts?

The two things I say on the regulatory front… One, the quality of regulation is only as good as the quality of the people in the regulatory context and in the government context. And so we have been relentlessly focused on the quality of people we have thinking about this question. The AI Security Institute in the UK is unparalleled in the quality of its people and the extent of its access to models to be able to understand where the risks are. And that's a really fundamental starting point--good capability in the public sector beats the quality of regulation every single time.

The second aspect of it is we have said where there is, in a sector specific way, a risk, or where there is an acute risk, let's say child abuse material, for example, we are not going to compromise on those questions. And so we have gone very robustly in both sector context and through aspects like the Online Safety Act, to say where the risks are here, we will regulate them. Where the risks are hypothetical or potential or prospective, we will build capability, we will dynamically move, and, yes, if needed, we will move in the direction of stopping those risks.

The UK is positioning itself to be an AI superpower. From that standpoint, what is your strategy?

We had an AI opportunities action plan over a year ago, and we announced that even just within a year, we've achieved 75% of it. I want to try and close the final 25% very soon. And so in terms of the UK's positioning as an AI superpower, I think there are three things that are really top of mind. One, being right at the frontier. Google DeepMind started in London, so a huge number of the advances remain focused at the frontier, and we're building great models in science and drug discovery and materials discovery, making human progress grounded in the UK.

The second part is we want to be the best country in the world at adoption. We have a particular focus on what creates value for people in the country. And the thing I'm passionate about in the Indian context is the Prime Minister's focus on welfare for all, happiness for all. We share that deeply in the UK context, and we want to do that through the medium of adoption of AI.

Then the third way in which we're positioning globally is being very thoughtful about responsibility in the adoption of AI as well. We have a great capability in government, the AI Security Institute looks at that. But more broadly, we think we can get the social policy, the institutions right to be able to make sure AI benefits all of us.

There's concern about the concentration of power in AI in the US and China. Does that worry you? That some countries are getting left behind? How do you think they can catch up?

One thing that I think all of us are recognizing is that AI is increasingly the frontier for both economic power and hard material power as well technologically. What we're focused on is making sure we're deepening our alliances. The US has been a great partner for us. We have not just an economic agreement more generally, but a very specific technology prosperity deal as well, which has deepened the US and UK relationship. We want to keep doing more of that with India as well. And I think in that context, the UK is in a very distinct position because, by virtue of where we've arrived, which is a position of deep trust with the US, with India, with allies, I think we're now increasingly a safe harbour.

If you're a US company in a sensitive area developing technology, the UK is a totally trusted place now. You should come and scale in the UK, and many of them (these companies) have. If you're an Indian company and you're thinking, ‘How do I partner with others in a country that is trustworthy, that we share values with?’ The United Kingdom should be top of your list. In this evolving geopolitical moment, the UK's history of relationships and pragmatism in deepening those will serve us very well.

What is your view on sovereign AI? More countries are pouring billions into having GPUs within the country. What is the UK's approach to that?

Sovereignty is a very ambiguous term that people use. There are three things that I think are critical to it. One, have enough critical inputs for AI. So yes, we want to get GPUs in the UK, and if that means partnering with companies we trust, we will get that. The second is to diversify your engagement with companies as well. One of the things I'm excited about here in the Indian context, is working with companies in India, fantastic innovators coming to the UK as their first international expansion. And then the third thing that is really critical is building domestic capabilities. We have amazing chip design startups doing AI for inference. We have great science model builders doing real frontier work. And we have some great people doing applied AI work as well. We've got a great delegation that I encourage everyone to look at. We've got great companies here as well. Each of those steps to sovereignty have enough of the inputs, diversify those inputs, and then, yes, build domestic capability. We're walking up each of those rungs of the ladder.

Tell us a little bit more about the areas where you think that India can contribute in the area of AI and where the UK can contribute toward building a greater AI ecosystem in India?

India has an exceptional tech ecosystem from a talent point of view. That talent building models can be quickly and economically efficiently adopted. I've just been speaking to some of your founders. They're building models that are small models that can be deployed in the palms of people's hands rather than just large data centre clusters. In the opposite direction, we have a great university system. Four of the world's top 10 universities (are) in the UK, a very R&D-intensive AI ecosystem. We want to make sure that we're using that to spread opportunity for people right across India as well. There's some fantastic British universities out here already making sure that brilliant, energetic Indian talent gets the benefit of that support as well. So I think those are two specific areas.

Dia Rekhi contributed to this story.