Ian Goodfellow, Inventor of GANs & Research Scientist at Google DeepMind
Insights from an AI Pioneer
It was such a great honor to have Ian Goodfellow join us at SV Icons for dinner this Saturday. Ian invented GANs, is a pioneer in AI, and currently works as a Research Scientist at Google DeepMind. Hearing Ian’s perspectives on a wide array of topics, from AI to conscious will to career lessons, was an amazing experience. Thank you so much, Ian! Thanks for hosting, Aizada Marat and Chyngyz Dzhumanazarov. A few key takeaways on non-confidential topics below:
How to Generate Interesting Ideas?
· Put yourself in a group or institution where people around you are each working to solve many different, interesting problems.
· If you are trying to solve a problem but cannot figure it out, pinpoint the exact reason why and monitor new technological developments that could potentially enable you to solve the initial challenges months/years later.
· Synthesize different ideas: Ian came up with GANs at a bar when a couple friends presented a problem to him (training a generator network to produce images), and he combined it with an existing idea (using a discriminator network to score a speech synthesis contest) to realize that a discriminator could help a generator make images by including it in the learning process.
What is A Key Lesson from Ian’s Career?
· If you think you can see what is coming next, go with your gut even if the respected people in your field disagree with it.
Do Humans Have Conscious Will?
· No, we have the illusion of conscious will, and neuroscience research shows we come up with narratives after the fact to explain our actions to ourselves.
What is One of Ian’s Favorite Podcasts?
· Check out Our Fake History (https://ourfakehistory.com)