A conversation with Bill Gates and Tara Westover

Bill Gates certainly needs no introduction, especially to the 50 students in the audience who are enrolled in our CNN Academy in Abu Dhabi. Today, these aspiring journalists are interviewing this world-leading technologist and philanthropist for his insights on how artificial intelligence can tackle major global challenges. Ahead of the CNN Academy Masterclass, we asked our students here to submit questions to you. One of the most consistent questions that we received was this, with your decades at the forefront of technological transformation, how surprised are you by the rapid development of AI at this point? And so many people have asked us about where you stand on the threats and opportunities of AI's growth. Let's start there. Well, the dream of AI goes back to even before I was born. Alan Turing raised the question, would there ever be computers that you wouldn't even know whether you were talking to a human or a computer? And for a long time, AI couldn't even do basic things like visual recognition and speech recognition. And as we scaled up and used these neural net techniques, one by one, we solved different problems. The computer got better at recognizing speech, got better at vision, really matching human performance in sensory areas. But the threshold we had not crossed that I often raised was it essentially couldn't read and it couldn't write. Reading means encoding the knowledge in a large body of text in a way that you can access it. Say, passing a medical test is a great benchmark of that. And that's where CHAT GPT-4 from OpenAI, just a little bit over two years ago, really took that unsolved problem and said, wow, this technique actually can give us pretty phenomenal results there. And in these two years, not only has OpenAI dramatically improved their models, but you have many others. So yes, I always expected this to happen, but it was going, the hardest problem, we weren't sure and then all of a sudden we had the breakthrough and now we're having some additional breakthroughs, what we call metacognition. And so it's the most profound advance in technology ever. Bill has been an outspoken advocate of AI on CNN and off. Leilani is in the audience. Please stand up and what's your question? As AI reshapes industries and economies, what steps are needed to be taken to ensure that under-resourced regions are able to also benefit, especially in healthcare, where that digital infrastructure maybe just isn't quite there? Typically, big advances become available to the rich countries and then it's often a generation or more, greater than 20 years, before they're widely available in middle income and low income areas. And if we're not careful, that's the same thing would happen here because although the basic interface, we can teach these AIs all human languages, not just written but also speech in and speech out. So gathering the data for all the languages of the world, including in the case of the Gates Foundation, the African languages, that's going on and so we'll at least have that part will essentially be solved. Getting the cloud computing resources that these intelligence services require available to those countries, there's a lot of great discussion about that. In health, now things like diagnostics, a lot of those tools are getting cheap enough that we can give those out. Even some diagnostics, you can use the camera on the phone or some simple add-on. We're trying to make ultrasound very cheap. We have now a thing where we use AI where a pregnant woman can be scanned and told, okay, your pregnancy will be a risky pregnancy so you ought to go to where a C-section is available or no, it looks like it's going to be fine. And that alone would have a dramatic benefit. So only by consciously coming up with the right data, supporting their languages, working with their regulatory apparatus to get comfortable and then having essentially donations of some percentage, even 4% of the cloud capacity would cover these pretty basic things. So I think we can make this an exception and get it out there almost at the same time. Beyond investing in this portable ultrasound from the Butterfly Network, the Gates Foundation has committed an annual sum of $370 million from 2023 to 2027 to support innovative solutions for maternal, newborn and child health. On a broader basis, where do you see or what do you most focus on with regard to AI and drug discovery? There's many of the world's diseases. Take malaria or tuberculosis as an example. Where the people who suffer aren't rich enough to command the attention in the marketplace. And so even though to this day over 600,000 kids die every year, 600,000 of the 5 million kids who die under the age of 5 die because of malaria. And yet malaria research is tiny. We're by far the biggest in that. And so now we're using AI. We're coming up with new drugs, a new vaccine, also a genetic way of massively killing mosquitoes, so-called gene drive, which will mean the infection rate goes down and give us a chance as all these tools come together to actually eradicate malaria. As optimistic as you've ever been about the eradication of malaria at this point then? Well, in terms of the upstream work, that tools discovery and the great progress we're making there, yes, that is fantastic. The world that we're delivering these tools into, instability in the Sahel, the distraction of terrible things in Sudan, Gaza, Ukraine, this is a world with a lot of tumultuous things and it means that the basic needs of the poorest countries, including in Africa, sort of we get distracted. The budgets go to the military, to the refugees, to a wide variety of things. And many of those that we are talking about affected by, for example, malaria, are in some of the poorest countries or poorest parts of the world, which are climate-inflicted. I mean, very specifically, if not conflict, then climate and oft times conflict and climate at the same time. And climate came up in so many of the questions that our students had for you. How can we harness AI in the fight to preserve the planet whilst also being aware of the massive appetite, of course, that AI has for energy? How do you balance those? Actually, AI is small compared to electrifying cars, electrifying trucks and buses and taking industrial processes like steel or cement and having electricity rather than direct hydrocarbon usage be a key input to those. But AI will add significant electricity demand. Now, fortunately, the companies that are involved in building these data centres don't want to add to climate problems. And so they're very desirous to find either nuclear fission or fusion or wind or solar type approaches. And so they'll be willing to invest and pay a little bit above, say, the natural gas price of electricity and drive some of these technologies down to be less and less expensive. And the appetite, of course, for that energy will be sooner than that. Oh, absolutely. In the near term, we have to build a lot of solar, unfortunately, probably some additional natural gas to meet the demands over the next six years. And you have said, and you've been consistent about this, that there will be a mix of energy provision and that will include traditional fossil fuels for a period of time, correct? Anyone who thinks we can make an abrupt transition just isn't being realistic. We need to make nuclear a lot less expensive. And we need to explore ways that you can take things like natural gas and use it without any leakage and potentially even ways of using it so there's no carbon dioxide.