Why Britain’s AI minister isn’t a tech bro

Is AI minister Viscount Camrose, Jonathan Berry, a tech bro? He laughs hysterically: “I don’t think anybody is going to be calling me bro!” 

The hereditary peer, the fifth Viscount Camrose, 53, may not claim tech bro status but boasts a long-standing interest in AI, having taken courses in the subject during his masters degree at Pennsylvania’s Carnegie Mellon University and then at Birkbeck, University of London. 

“I’ve been interested in AI, actually, since I was quite a small boy. My father, Adrian Berry, who was a science writer at The Telegraph, wrote a book [The Super-Intelligent Machine] about it, which I must have read when I was about 11,” Viscount Camrose says.  

“He, like me, was very, very keen on the film 2001: A Space Odyssey, which is the greatest film and is still my favourite film. I was very, very interested in the Hal 9000 computer and I used to love the way that it could do things that none of the humans expected it would be able to do like lip reading.” 

One of those unexpected things was when the Hal 9000 tried to kill the crew. Let’s hope that as AI minister, he doesn’t let it get to the stage. 

When Viscount Camrose entered the House of Lords, on his second attempt in the elections that allow (currently 91) hereditary peers to continue to sit in Parliament’s upper chamber, he made sure to join plenty of technology All Party Parliamentary Groups (APPGs) and committees. It was less than a year later that –  to his “delight and amazement” – he received a call from the Chief Whip asking him to become AI minister. 

“If there had been any whisperings of it, they had certainly passed me by. It was completely out of the blue,” the AI minister chortles. “I pretended to think about it for two days.” 

It seems to be his career in tech management consultancy that caught Rishi Sunak’s attention, or more likely, Viscount Camrose says, No 10 approached the Lords looking for a suggestion of someone clued up to take on the brief. 

He has been using AI tools to help with the ministerial job, for example, asking the generative AI chatbot ChatGPT to summarise the Online Safety Bill for him: “What’s so brilliant with ChatGPT and the summarisation is that it is so much better than people at taking huge chunks of text and giving me a summary.” 

Viscount Camrose adds: “That’s very helpful. Particularly in this job where the key is being able to absorb huge amounts of information really quickly.” 

The great thing about it is to find ways that AIs dynamically align their goals to human needs. That is the trick

The AI minister, alongside his DSIT colleagues, was recently turned into a 90s version of himself using generative AI images: “It was a little bit cringy but it was quite funny. I mean, all of these things in their own way, I think they tread a line between fun and amazing, and serious and scary.” 

The serious and scary side that he refers to surrounds deepfakes that could be deployed to disrupt politics. “I’m told by psychologists that there’s a real danger here,” Viscount Camrose says. “There was a really nasty one about the leader of the opposition [a deepfake audio was published that made out to be Labour leader Keir Starmer abusing his staff and criticising the city of Liverpool]. What it does is there is an anchoring bias  so if you get presented this, even though you’re told it’s not true, that’s your starting point when you think about the person. It’s very, very alarming.” 

Does he believe the next election will be an AI deepfake election? “It is a risk we take incredibly seriously,” the AI minister says. 

While he recognises that you could see fake news before, what we now get, Viscount Camrose adds, is “micro-targeted, fake news” that makes it “very difficult for you to identify”. 

The AI minister says: “There are a lot of other Western democracies having elections between now and whenever our election ends up being that we are able to observe and see what happens. But that is, I think, the most immediate risk.” 

He adds: “We are going after that in a very big way.” 

But Viscount Camrose won’t be drawn on the details of exactly how the government is going to do that. He flags that there is already the Counter Disinformation Unit (CDU) working within government to “keep a lookout”, “make rebuttals” and “engage with the big social media platforms”, so that in the case of the deepfake against Keir Starmer, the CDU could make a public rebuttal that says: “You may have seen something that appears to have been said by the leader of the opposition. This is not the case.” Although that doesn’t appear to have happened this time around. 

A number of parliamentarians have expressed their concerns about the CDU, with a group of cross-party MPs writing an open letter to the government claiming it is collecting examples of legitimate posts criticising policy. 

“I really understand where people are coming from, that we don’t want the unit of government out there correcting communications, but we have to in the world we’re in,” Viscount Camrose says. “Disinformation is being weaponized against us and we have to respond. Obviously, freedom of speech is the absolute non-negotiable condition. But we have to respond to its misuse.” 

He is planning on sitting down with members of the House of Lords to give an operational briefing of exactly how the CDU works. 

Another area of concern that has become inflamed recently is over AI facial recognition, with another group of MPs and peers calling for an “immediate stop” to the practice.  

Viscount Camrose says that when it comes to facial recognition, “context is everything”. 

“I do not want facial recognition to be used for certain things like marketing, in most cases, policing or trying to shape my behaviour. Absolutely no way, no how, I am very firmly against that,” he adds.  

Perhaps putting himself in opposition to policing minister Chris Philp who announced government plans to make United Kingdom passport photos searchable by police, integrating data from the police national database, passport office and other national databases to help police find a match with the “click of one button”. 

The AI minister adds: “There are a lot of perfectly benign uses of facial recognition, like logging into your iPhone. We can imagine a situation in which a dangerous terrorist was on the loose, facial recognition would be quite handy. 

“It is about the context. In certain contexts, I would want facial recognition to be used, in the vast majority of contexts, I wouldn’t want to be anywhere near it.” 

Despite the negatives, Viscount Camrose maintains his position as “an optimistic realist”: “I genuinely come from a place of deep optimism, which is not to say, let’s all be naive about the risks. The risks are very, very substantial, both the shorter term and the longer term, more existential stuff, and we have to get those right. But there’s a massive reward for being able to do so. Which I think goes to address an awful lot of problems not just in UK society, but in the world.” 

Government relationships with big AI labs like Google DeepMind, Anthropic and OpenAI, are “part of the solution”, the AI minister says, though “not the whole of the solution”. 

“It goes without saying, if you only talked to very large AI labs, you would have a strong set of views, but in one direction.” 

The government’s Frontier AI Taskforce, set up to examine the risks of AI, has been in talks with such companies to gain access to their models to better inform government understanding. Has that access happened yet? 

“I think the exact status of those is rapidly moving and ongoing,” Viscount Camrose says, “all building up to the summit”. 

Is the aim in that case to have access by the start of the AI safety summit in Bletchley Park this November? The AI minister gives his strongest hint that will be the case: “Let’s say, the aim is that the role of the private AI labs will be clarified by the summit.” 

Viscount Camrose’s ambition for the summit, he says, comes in layers: “Think of it like a cake. The base layer is producing a shared statement of the risks because I think that gives us a lot of difficult definitions. What is AI? What are the risks? What are the benefits? A shared philosophical agreement on all of those things is not trivial but that’s the minimum I think we need to have.  

“The layer above that is what should we going forward as governments, as creators of AI, as companies, be doing about this differently? It’s not going to answer that question but it’s going to put in place the steps that get us to the answer. The third thing, which I think is really very important too, [and] sits on top, is a demonstration of AI as a force for good.” 

How does China fit into this? They have been invited to the AI summit despite criticisms from some MPs. Not inviting them, the AI minister says, means you would be okay with having “a bipolar world in terms of frontier AI risk”. 

He adds: “It’s not about collaborating around innovation. It’s about collaborating around risks that can impact all human beings. We all have the same stake here.” 

Reports are rife that China will be involved in the first day of the summit but excluded from key talks on the second day. Is that the case? 

“There are different activities for different people on both days,” Viscount Camrose says. 

In certain contexts, I would want facial recognition to be used, in the vast majority of contexts, I wouldn’t want to be anywhere near it

What does he make of the dominating theory that artificial general intelligence (AGI), so-called ”God-like AI”, could complete and even surpass intellectual tasks humans could perform? 

“Ten years ago, if you’d gone around and asked computer scientists, when would we get to AGI? The answer would have been somewhere between 50 years and never,” the AI minister says. “Now, you know, that number has gone way down. It is a very serious risk. Most people, I think, believe it’s solvable.” 

He includes himself in that group: “The great thing about it is to find ways that AIs dynamically align their goals to human needs. That is the trick of it.” 

Speaking to the BBC, Viscount Camrose once remarked that AI “makes everybody massively more productive whatever it is you do”, almost making you into “the Marvel superhero version of yourself”. Is that really the case for everyone? 

The AI minister recognises that he was perhaps overzealous in his previous comments and clarifies: “I think we could find jobs that it does not make more productive. But not as many as you would think.” 

In May, IBM put the brakes on nearly 7,800 jobs that could be replaced by AI and automation over time, shortly before BT announced it is predicting to replace around 10,000 jobs by 2030 with AI. 

The AI minister’s example is along the more creative side. If you were a visual artist, he says, “I can absolutely see that you’re worried that your work is going to be taken.” 

“On the whole,” however, “for the great majority of jobs, I’m convinced it does give you this productivity. If you’re a delivery driver or a postman, they can actually reorganise your route more efficiently, improve your carbon footprint.” 

Viscount Camrose has been advocating to other members of the Lords the benefits of AI, in conjunction with their usual work. 

One Lord, he says, approached him to say that he had decided not to hire a new researcher in favour of paying a small fee to use ChatGPT. “The debate we had was me saying: ‘Well, maybe that’s not the right way to look at it. Hire the researcher but have the researcher then working one level of abstraction up with AI, where they can manage lots of different new research projects, rather than focus on one thing [in] great detail.’” 

The AI minister says: “That is, I think, the way forward.” 

PoliticsHome Newsletters

Get the inside track on what MPs and Peers are talking about. Sign up to The House's morning email for the latest insight and reaction from Parliamentarians, policy-makers and organisations.