Demis Hassabis

Discovery

BBC

Science, Technology

4.3 • 1.2K Ratings

🗓️ 10 March 2020

⏱️ 27 minutes

🔗️ Recording | iTunes | RSS

🧾️ Download transcript

Summary

Jim Al-Khalili finds out why Demis Hassabis wants to create artificial intelligence and use it to help humanity. Thinking about how to win at chess when he was a boy got Demis thinking about the process of thinking itself. Being able to program his first computer (a Sinclair Spectrum) felt miraculous. In computer chess, his two passions were combined. And a lifelong ambition to create artificial intelligence was born. Demis studied computer science at Cambridge and then worked in the computer games industry for many years. Games, he says, are the ideal testing ground for AI. Then, thinking memory and imagination were aspects of the human mind that would be a necessary part of any artificially intelligent system, he studied neuroscience for a PhD. He set up DeepMind in 2010 and pioneered a new approach to creating artificial intelligence, based on deep learning and built-in rewards for making good decisions. Four years later, DeepMind was sold to Google for £400 million. The company’s landmark creation, Alpha Go stunned the world when it defeated the world Go champion in South Korea in 2016. Their AI system, AlphaZero taught itself to play chess from scratch. After playing against itself for just four hours, it was the best chess computer in the world. (Humans had been defeated long ago). Many fear both the supreme intelligence and the stupidity of AI. Demis imagines a future in which computers and humans put their brains together to try and understand the world. His algorithms have inspired humans to raise their game, when playing Go and chess. Now, he hopes that AI might do the same for scientific research. Perhaps the next Nobel Prize will be shared between a human and AI?

Transcript

Click on a timestamp to play from that location

0:00.0	You're about to listen to a BBC podcast and trust me you'll get there in a moment but if you're a comedy fan
0:05.2	I'd really like to tell you a bit about what we do. I'm Julie Mackenzie and I commission comedy
0:10.1	podcast at the BBC. It's a bit of a dream job really. Comedy is a bit of a dream job really.
0:13.0	Comedy is a fantastic joyous thing to do because really you're making people laugh,
0:18.0	making people's days a bit better, helping them process, all manner of things.
0:22.0	But you know, I also know that comedy is really
0:24.3	subjective and everyone has different tastes. So we've got a huge range of comedy on offer from
0:29.8	satire to silly, shocking to soothing, profound to just general pratting about.
0:35.0	So if you fancy a laugh, find your next comedy at BBC Sounds.
0:40.0	This is Discovery from the BBC. I'm Jimel Killelli and in today's program I'm in conversation with a leading scientist about their life and research. Welcome to the life scientific.
0:52.0	Demes Hasabes is the brains behind some of the most sophisticated artificial intelligence systems on the planet.
0:59.0	Demes knows more about how machines think than anyone I know and his company Deep Mind was sold to
1:05.1	Google for 400 million pounds. He retired from international chess
1:10.5	championships at the age of 12, worked in the computer games industry for many
1:14.9	years and later studied computer science and neuroscience. Using the things he
1:20.0	learns from all these different areas of interest he learned a
1:23.4	a radical new approach to creating artificial intelligence.
1:26.6	His computer systems are not pre-programmed, they teach themselves.
1:31.6	His AI system, Alpha Alpha Zero defeated the best chess computer
1:35.8	in the world having learned how to play chess from scratch in just four hours.
1:40.9	Now Demis wants to use AI to help humanity and he hopes one day humans and
1:47.0	AI working together will make a Nobel Prize winning discovery.
	...

Please login to see the full transcript.

Previous episode | Next episode

Disclaimer: The podcast and artwork embedded on this page are from BBC, and are the property of its owner and not affiliated with or endorsed by Tapesearch.

Generated transcripts are the property of BBC and are distributed freely under the Fair Use doctrine. Transcripts generated by Tapesearch are not guaranteed to be accurate.