It's an exciting day for us here at the World Government Summit, talking about AI, talking about our future — something that everyone here, no doubt, is incredibly invested in. Lila, I guess the first question for you is: how worried should we really be?
I think this is a very important time in AI's development. We're actually still in the early stages, which means that we're never going to have as much time as we have now to experiment. We're never going to have as much time as now to shape how the technology gets developed, how it gets deployed. And so I actually think, you know, part of my job is both to worry, but also to make sure that we're investing in the opportunity. I think, as mentioned before, AI has the potential to help us solve some of humanity's biggest challenges, but that really requires us being engaged.
Yeah. How do you feel about the current ecosystem? Because there are a lot of fears, certainly in the media, but I think they're fears that all of us share in terms of the job loss potential. In your mind, are the governments prepared?
Well, you know, I think we've seen in the past how technology has really shifted. I've been in the technology industry for over 30 years, and just seeing, you know, I remember the days even back in computing and internet when that was on the rise, there was a shift in the job sector. And I think we can take a lot to learn from the past, but also I think AI has a potential — it's already getting more general and more capable. And so I think now is the right time for us to be working more closely with governments. So one thing we are seeing is here in the UAE, and where I work in the UK, governments are really starting to lean in to say, how do we use AI as a way to help address some of the bottlenecks, whether it's, you know, public planning and trying to streamline the process.
I was just in Northern Ireland, where teachers have been experimenting with Gemini, and they've actually saved 10 hours per week per teacher and are able to serve students with different learning needs. So I think we're already seeing where this is making progress. Now, AI is such a transformational technology. It will require exceptional care, and that means we have to regulate, but we have to regulate well. We have to be able to mitigate, manage and mitigate the risk, which requires foresight, and we also have to make sure that we're supporting the innovation for entrepreneurs to be able to innovate, for us to use AI in healthcare to get better healthcare, deliver better services. So I think it's this: how do we manage it and collaborate in the right way.
For those leaving university right now, there are a lot of fears that already perhaps their degrees are obsolete. In your mind, what does education as an industry need to do to step up?
Yeah. You know, this is actually an area I'm particularly passionate about. Listen, AI is not going away. And the more we just try to layer AI on top of existing infrastructure, we're going to start seeing the challenges and the bottlenecks. And one area I'm actually particularly concerned about is if we're not supporting responsible usage and teaching that at a young age, both with professors or teachers and students, then how do we expect them to have a role in shaping the technology?
And when you say responsible usage, is that just asking silly questions and wasting the energy? [laughter]
No. I'll give you an example. I was talking to a teacher the other day, and they're putting together an AI student council for their school, such that, you know, one teacher says you can use AI this way, another teacher says don't use AI, and what's happening is the kids are using AI, the teachers just don't — they can say whatever they want. So instead, what they're trying to do is flip it and say, let's get the students working with the teachers to put together a framework that we agree on within a school, so at least there's some general guidelines. It's kind of like in your office environment too. You really want to set the ground rules, because if you don't, you have inequitable use and you start to actually accelerate a gap, rather than thinking about how are you responsibly going to do this and learn from it, so that you can change the guardrails, you can inform what's needed from a safety or risk perspective, and also highlight where things are working to accelerate.
So just a few weeks ago, Jamie Dimon suggested at Davos that governments are going to have to create incentives for financial institutions and corporates to keep people from being shed at an alarming rate as AI becomes more and more a part of business. In your mind, what do governments need to do now to prepare for this?
Well, I think the first thing is: how are you supporting, how are you using AI? So the first step is for governments to actually — individuals to start using it. Now, sometimes that might be in your work, but it's also in your personal life. AI is a general-use technology, and our belief is that it actually should enhance human capability, not replace it. And so I think the first thing is: experiment with the tools. Again, you're never going to have as much time as you have now to come up to speed.
And the other thing is, there are a lot of really interesting use cases around the world, and I've been particularly impressed with some of the conversations we've had here in the UAE about really leaning into it. I mean, look, it's the first government with an AI minister, the first country that has an AI university. There's some really interesting applications the entrepreneurs are coming up with. You've lowered the barriers. You no longer need to code at the same level. So you can start ideating before you build a company, removing some of the barriers that existed before. I think healthcare, understanding diseases, addressing it. I think what's exciting about this is we're seeing the usage and the impact. And so it's more of a matter of how do we bring that to fruition and support the growth.
Tell me this. We've all heard the stories of AI bots essentially, or AI getting ahead of itself and speaking — these AI agents are speaking to each other now. They don't need the humans. They're talking about the humans. And they don't necessarily have great things to say about us. In your mind, is this something — you know, this used to be science fiction and now it's actually happening today? Are AI companies ahead of this?
Well, so maybe first a quick background. DeepMind was founded in 2010. So well before, you know, we're seen a lot as one of the leading labs within the world to bring about the modern era of AI. And I have to say that even from the beginning, the founders started this with a concept of, like, we have to build responsibility in from the beginning. And what that means is research. So we've done a lot of work around ethics research. Before we did our work on agentic technology and published a paper or did a technology demonstration, we actually did a lot of research around the ethics of advanced AI assistance. We didn't have a vocabulary to talk about the risks or the benefits. So we do a lot of that investment upfront, because responsibility or safety checks shouldn't be a bolt-on. Cyber security shouldn't be a bolt-on. You shouldn't wait till the end. For research to really flourish, it needs to happen from the beginning and hand in hand. So what we're seeing is by building an interdisciplinary team, by facilitating — as leaders, I think it's our job to facilitate the conversation so that we can identify that and do some of the mitigations upfront.
Does it worry you specifically?
