AI Transformation and Cybersecurity Challenges

ACAS powers the world's best podcasts Here's a show that we recommend Ionica Reingel, nutritionist, author, and host of the Nutrition Diva podcast We dig into the questions that you're actually asking, if it's okay to drink coffee on an empty stomach, whether it's possible to retrain your sweet tooth, which ultra processed foods you might actually want to include in your diet we take a closer look at diet trends. Fact check sketchy claims and track down the science so that you can feel more confident about what's on your plate. New episodes are released every Wednesday Find Nutrition diva on Apple podcasts, Spotify, or wherever you're listening, and be sure to follow or subscribe so you don't miss a single episode A cast helps creators launch, grow, and monetize their podcasts everywhere. Acast. com This episode of the Business is sponsored by PWC. Markets are shifting, teechnology and AI are reshaping industries. Disruption is unavoidable. staying ahead and finding new opportunities for growth demands bold moves This is where PWC makes things happen Working with you, combining the industry expertise, data insights, AI capabilities and business change experience. to help you reinvent and lead from the front PWC accelerates what's possible so you can turn vision into value. Discover more at pwc. coot u Hi, Hannah Prevt here Last week, I sat down with Kathleen Fisher, who's the Chief executive of the Advanced Research and Invention Agency Aria was the brainchild of the former Conservative Party advisor Dominic Cummings. and when it was set up three years ago, it was modelled on DAPA, America's Defense Advanced Research Pjects aggency where Kathleen used to work This conversation was her first since she joined Arya back in January Now interviews span topics like Britain's chances of creating a trillion dollar company How to use AI to empower rather than replace people. and her concerns that the UK is too risk adverse Here's a piece of the discussion which has been edited for length and clarity I started by asking her what lessons she had brought from her time at Darfa. whether or not she's going to make any changes to Arya So I actually led the Information Innovation offffice, which is the computer science part of DARPA. One of the things that I would highlight is the need to focus on the translation piece or the crossing the valley of death. So the scientists kind of like to think that when you've solved a science miracle, the progression into changing the world just happens automatically. And maybe over the span of twenty or thirty or forty years, that might happen but it doesn't happen automatically. And so focusing on how do we jump startart it? How do we make sure that translation happens? and how do we make it happen as fast as possible? A Dartby worked on impact maximization plans with the program managers who worked in my office. and we're adopting that same practice at Aria. We are in the process of hiring a chief translation officer, so there'll be an executive whose job is to work with the program directors at ARA to strategically be thinking from the very, very beginning of the program to if our science miracle comes true How are we going to get it to translate into real world as soon as possible already, AREA programs have a theory of change, where we don't start an AREA program without thinking about how is this going to benefit UK citizens and do we have a performer base or creator base in the UK to make this happen? Arya doesn't do long term commitment Eventually, Aria is going to leave and go do other projects. So who is going to take Aria's place to convert that science miracle into that societal impact miracle? How do we get those downstream people excited? If it's starting new companies, how are we going to get venture capitalists excited? Are there regulatory hurdles that we have to solve How do we get the regulators on board right away? And can we get those people involved from the very very beginning so that we know as soon as we solve the science miracles, we have all the other ducks lined up or the dominoes lined up so they can start falling as soon as possible So Cathleine, we're living through one of those periods where multiple technologies are advancing at absolutely breakneck speeds. How do we really choose where the UK can win I think there are many, many places where the UK can win because of the UK's amazing science and technology base and the universities, the fact that we have such amazing science and technology base. Whenever AREA creates a program, part of what we do is this competitive analysis to say like does the UK have a unfair competitive advantage in this space every program that RAA launches We require that eighty percent of the effort would be funded in the UK. So like AREA is expected to provide value to the UK in at least three different ways. The first is the value creation that I talked about. The second is the value captuure, the financial return of the investment being in the UK. And the third is funding the research itself in the UK, right? If you just look at AREA's portfolio E one of those efforts had an analysis done that said like the UK had an unfair competitive advantage in that space. And there's things like arm There's the pharmaceutical research. There's many, many areas, like the quantum, right? There's many spaces where the UK has a massive advantage And I suppose it's also thinking about what winning actually looks like when we are all talking about the anthropic IPO and Peter Kyle you spoke last September about creating a one trillion dollar company here in the UK. Can we build a one trillion dollar company in the UK could eventually. Yeah. I mean, winning doesn't have to necessarily be a trillion dollar company. You could have many, many, many half trillion dollar companies that could also be winning. In an AI space, you might not create another frontier company, but you could create a piece of technology that is absolutely necessary for the overall AI ecosystem R AriA has the scaling inference lab, which is designed to test new kinds of AI hardware that could become a necessary part of the overall ecosystem for building new kinds of AI hardware. We sort of know that the way we're building AI hardware historically is not going to be the long term answer because of realities of physics as you get really, really small, things start to not work anymore. What are you talking about when you talk about AI hardware? you talking about chips? GPUs, yeah. just like as you get smaller and smaller, you start to get quantum effects and the amount of power that is necessary becomes unsustainable. And so you start to need to explore different kinds of paradigms you're going to deliver the number of cycles necessary to get the amazing AI systems that we are seeing. And so you need to explore can we have hardware that leverages approximation instead of being super, super, precise kind of analog computing kinds of answers? That's a research frontier, like we don't know the answers there, but they offer better power kinds of answers And There's a whole industry of small startups that are trying out different pieces of novel hardware The problem is that those individual pieces of novel hardware, you can't evaluate, is that piece of hardware good without having a whole system wrapped around it? So the scaling inference lab is designed to be able to build a whole system so that individual companies can slot in their piece to be able to test, is that piece useful for the whole system So that's a place where the UK can provide a a resource that could potentially become indispensable for the overall industry So what's holding us back? because that all sounds glorious, but Where are those major obstacles and where is there still work to be done to overcome them Um Yeah, so I think One of the things we should think about that may be holding the UK back is just a mindset question. Like you can like think about, okay, we need to consider all the risks. And like there's this risk and there's this risk and there's risk and' this risk. But the risk of not going fast enough and the risk of not doing anything, it's really easy to just not weight those risks appropriately One of the risks that DARPa takes really seriously is going too slowly versus other funding agencies that spend forever doing discovery of related work. They spend so much time thinking about, has anybody else ever done this before that they miss opportunities to do anything. And you in computer science background assumptions change so fast that even if somebody had done this a year ago or two years ago or five years ago, It would have been different because the situation changes. So the risk of going slow is worse than the risk of maybe replicating the work before. So the risk of Not doing something can be much worse than the risk of maybe doing it not exactly in the right way and learning something from doing it. And are we particularly risk averse in the UK compared to the US, we often hear a lot about the attitudes of failure versus the US. I mean, I've only been here for three months and it's an easier stereotype, but had a couple of conversations with people about And How do people cope with like they go and start a company and it fails? Like how do they cope with that failure like that staying on their resume And it's like It's not seen as a stain on their resume. It was like it's kind of that a failure of a startup is like that's the expected outcome They learn stuff out of it. that's like a learning experience. And like, well, how do investors deal with their startup failed? And it's like, well, people who are investing in startups don't expect they don't invest in one startup. They invest in a family of startups and they expect one out of ten to succeed. And that one that succeeds covers the other nine that didn't succeed. And of course, you have to have expertise in how to invest in startups Again, that's a stereotype and I've been kind of unruck by the number of people who've talked to me about like that kind of mindset. So I do wonder if some of it is linked though to the fact that it's so much more difficult to raise capital in the UK compared to perhaps the US, particularly amongst underrepresented founders. But one of the biggest debates in Britain today, arguably is whether pension funds and other institutional investors should be allocating more capital to startups and scale ups. Do you think we're getting that balance right? Yeah, so I'm not sure. AriI's job is to get science miracles and then to get the translation miracles. And so we're working to create a lot more really great startups that would be ready to have investment for VCs. So hopefully we can sort out this question so that there will be more patient capital in the relatively near future becausecause I think there are going to be lots of really good companies coming down the pipeline. And there could be really fantastic win wins, both for investors and for the British people. And it would be great. if we do manage to solve the very significant miracles related to solving the common cold, it would be great if not just the value creation, but the value capture happened in the UK And much of the conversation, as it has this morning, tends to focus on kind of start upps and scale ups, but many of our guests here in the room This morning are from large established companies that will have huge sprawling legacy IT systems. You mentioned earlier and so did John that the biggest risk is doing nothing at all. And I've certainly spoken to CEO's who are just throwing a bunch of stuff at the wall and hoping something sticks. So what conversations should CEOs and boards be having that perhaps they're not having around AI implementation Yeah, I think that it is becoming increasingly clear that AI is going to be transformative across many, many different sectors. I kind of like think about that the AI is like a tidal wave where the wave has sucked out. if you think about the beach where the water is no longer there and the wave is out, and it's going to be crashing in. and organizations that figure out how to make use of it are going to kind of ride the wave as it comes in and organizations that don't are going to drown as the wave comes in and crashes I think organizations that are newer have an advantage of their systems are still pretty plastic and they can innovate really quickly. and organizations that are older have more challenges because they have legacy systems