{"url": "https://www.youtube.com/watch?v=RsFHe3b3nJc", "page_url": "https://www.youtube.com/watch?v=RsFHe3b3nJc", "title": "Mustafa Suleyman | The Ethics of A.I.", "show_or_channel": "The Telegraph", "date": "2018-03-21", "duration_s": 533, "type": "interview", "speakers": ["Mustafa Suleyman"], "confidence": "medium", "notes": "The Telegraph short interview on AI ethics (2018, ~8 min). Included as a standalone item from the DeepMind period.", "people": ["mustafa-suleyman"], "id": "2018-03-21_the-telegraph-mustafa-suleyman-the-ethics-of-a-i", "audio_duration_s": 533.056, "worker": "jc-w18", "transcribe_s": 24.0, "model": "large-v3", "segments": [{"start": 0.46, "end": 29.3, "text": "Just what is artificial intelligence? Well, most people would agree that AI, as it's known, is the technology that allows machines to learn and perform complex tasks efficiently and creatively. It's the buzzword on the lips of every tech company these days. And from personal assistants, hi Siri, to behind-the-scenes algorithms that can see patterns in vast and formless reams of data, it's already in all our lives. But where is AI going to lead us?"}, {"start": 29.3, "end": 58.88, "text": "And who is leading that charge? Let's take a look at some numbers. Around the world, it's estimated that technology companies spent about $30 billion on artificial intelligence research in 2016. And latest figures suggest that there are more than 250 AI focused companies in the UK alone, researching everything from cybersecurity to personal health. Other companies are using AI to improve efficiencies in things like food delivery. Even the tax man is getting in on the act."}, {"start": 58.88, "end": 88.86, "text": "According to a major government review last year, AI could add an additional \u00a3630 billion to the UK economy by 2035. People may be more worried about losing their jobs to smart machines than the rise of the Terminator, but fears about super-intelligent robots are not just sci-fi paranoia. Some of the tech industry's biggest personalities have spoken out against the potential risk of unregulated AI. I'm very close to the cutting edge in AI, and it scares the hell out of me."}, {"start": 88.86, "end": 117.49, "text": "I think the development of full allegiance spells the end of the human race. Earlier this year, Prime Minister Theresa May even announced the creation of a National Centre for Data Ethics, acknowledging concerns that this rapidly developing technology could one day end up making decisions without human oversight or approval. Britain is playing a crucial role in this emerging technology, and if we get it right, it should have profoundly positive consequences."}, {"start": 121.84, "end": 150.74, "text": "So I've come to one of the most powerful and familiar names in the tech world, Google, to look inside its controversial artificial intelligence unit, DeepMind. Now, DeepMind was founded in 2009. It used to be a British company, but then in 2014, it was acquired by Google for $500 million. The unit says its mission is to solve intelligence, and it's making fast work of that. In 2015, its algorithms became the first"}, {"start": 150.74, "end": 179.06, "text": "To take on and beat a professional player of the fiendishly complex board game, Go. And then last year, it went one better when the AlphaGo Zero algorithm learnt the rules of Go for itself and took on and beat the world's best player of the game. Obviously, the rewards of such power and such technology are immense, but so are the risks. And that is what preoccupies one of DeepMind's three founders, Mustafa Suleiman."}, {"start": 179.6, "end": 205.74, "text": "So humanity took 2,000 years to get as good as it possibly could at Go, and your algorithm took four hours to beat the world's best player. That's right. I mean, it's pretty striking. I mean, just to kind of rein that in a little bit, unfortunately, we can't quite yet use those sorts of methods in the real world. So the particular characteristics of the training environment that we had for Go are fairly unique in that we can..."}, {"start": 205.74, "end": 231.2, "text": "Define the rules very clearly and we know exactly what success or reward looks like. In the real world, it's actually very, very difficult to define your problem in a similar kind of way and you can't train your algorithms through self-play. What are the applications that you're looking at to help people to change humanity? The really exciting thing about these algorithms is that they are inherently general."}, {"start": 231.2, "end": 257.09, "text": "The same kind of algorithms that we develop to play the game of Go or to learn to play chess against itself. We also use to control and optimize Google's data center fleet, for example. Last year we managed to deploy an algorithm that reduced the amount of cooling energy used to cool down Google's data center fleet by 40%. Another example would be in healthcare. So we've actually trained"}, {"start": 257.09, "end": 285.22, "text": "An algorithm to detect a whole series of potentially blinding diseases from eye scans. I know you wrestle with these questions all the time. So how do you answer? They're not no one's, Musk and people like Hawking. How do you respond to them and reassure people? Look, I think with any kind of technology, it's easy to speculate philosophically about the very long term potential development course that a certain technology can take."}, {"start": 285.22, "end": 308.05, "text": "And that's quite fun and interesting, and it's easy to reference the sci-fi. I'm much more concerned about the practical near-term social and ethical consequences that we're going to see from complete digitization over the next decade. Many of our industries and sectors are headed towards digitization unlike anything we've seen before. And that is going to deliver tremendous benefit."}, {"start": 308.05, "end": 328.11, "text": "And I'm really excited about the utility that they can bring. But at the same time, I think it raises a whole series of interesting ethical questions, which we as creators of those technologies care very deeply about. In fact, it's the core reason why we started the company, because we want to deliver safe and ethical AGI in the world. And that's what we've been focused on."}, {"start": 328.11, "end": 358.05, "text": "I was the founding co-chair of an organization called the Partnership on Artificial Intelligence, where we brought together all of the major technology companies, Microsoft, IBM, Apple, Amazon, Facebook, who are working on artificial intelligence, and scores of other nonprofit organizations, academics, policymakers, to start a new completely independent organization looking at the social and ethical consequences of AI. And we think that we urgently need to have that conversation."}, {"start": 358.05, "end": 374.77, "text": "Draw kind of philosophical speculative stuff away from the kind of long-term superintelligence conversation and focus much more on the practical near-term ethical and privacy issues that I think are very pressing. Now is the time for us to prioritize."}, {"start": 374.77, "end": 404.11, "text": "Public engagement, prioritize hearing the voices of many communities who I think feel unrepresented or marginalized in the conversations about how our technologies impact the real world. We put a lot of effort into trying to stay in London. We believe that London is the kind of diverse, multicultural, modern city with many, many different industries and sectors, including the public sector and the third sector. And that environment, those values that come through in London."}, {"start": 404.11, "end": 433.34, "text": "The UK have really shaped the kind of company that we've built. And I think that's quite different to other places in the world that we could have chosen to build the company. And that's really important. Like Silicon Valley. Exactly. I think Silicon Valley, as we've seen much of last year, is sort of finally waking up to the reality that you cannot build technologies in a vacuum. So you're not always optimistic. You look forward and the future is, I suppose, unclear to all of us."}, {"start": 433.34, "end": 447.2, "text": "I would go as far as to say I'm not by default optimistic. Civil rights didn't happen by accident. They happened because of collective struggle. They happened because people stood up and said,"}, {"start": 447.2, "end": 474.51, "text": "This does not work for this marginalised group of people. And I think the same kind of conversation is going to have to happen if all of the technologies that we're developing, not just AI, but the entire digital revolution, if they are to be in service of the majority of people. That's why I started the company, that's why I sold the company to Google, and that's why I continue to run our division, because they're the things that I care about first and foremost. Mr. Solomon, thank you so much. Great to see you."}, {"start": 474.51, "end": 503.63, "text": "It's so exciting that Britain is such a pioneer in this emerging technology of artificial intelligence, something that is sure to change all our lives. In all my discussions with tech experts around the country, it's the names of Mustafa and his DeepMind co-founder, Demis Hassabis, which keep coming up as our nation's real game changers. Because just as mobile was at the heart of what's being called the fourth industrial revolution, so artificial intelligence is central to what's already being called the fifth."}, {"start": 503.66, "end": 522.06, "text": "So will humanity be freed to live a life of infinite leisure or have we sowed the seeds of our own destruction? Reassuringly, Britain's tech pioneers say they know the risks as well as the rewards of artificial intelligence. Either way, Britain is sure to be at the forefront of what's to come."}]}