,
11 tweets,
2 min read
Read on Twitter
A thought-provoking article by @GaryMarcus . Accordingly, I had some thoughts! Longish thread ahead. /1 wired.com/story/deepmind…
First thought: Why frame this as "DeepMind lost $572 million" rather than "Google made a very large, long-term investment in one of its research arms, DeepMind"? /2
DeepMind's mission is to "solve intelligence" -- a statement I find nonsensical BTW -- but whatever you want to call it, AI is a very hard, long-term research problem and it's great for companies to fund basic, long-term research that doesn't have immediate payoff. /3
Second thought: Marcus claims that deep reinforcement learning is "a kind of turbocharged memorization". That's an interesting but still too vague hypothesis that I think needs to be made more formal and testable. /4
Third thought: Marcus: "DeepMind has yet to find any large-scale commercial application of deep reinforcement learning." Why the focus on this kind of short-term commercial application? Was that kind of short-term payoff what Google had in mind when it acquired Deep Mind? /5
Fourth thought: Marcus compares DeepMind with IBM Watson. But IBM made huge promises & put out a lot of hype on how their system would very soon revolutionize healthcare, law, etc. Has DeepMind ever made promises about how it would commercialize Deep RL in the short-term? /6
Fifth thought: Marcus correctly warns of the dangers of "overpromising". But is this fair, with respect to Deep Mind? Have they actually overpromised with respect to their technology? (I did see a quote from Shane Legg a while ago personally predicting AGI by 2020s 🙂) 7/
Final thought: I completely agree that AI research should pay more attention to cognitive science.... 8/
But I'm not sure about the glib comparison Marcus (and a lot of other people) makes: "How do children acquire language and come to understand the world, using less power and data than current AI systems do?” 9/
I suspect that it's really hard to assess the amount of "power" and "data" used for a child to acquire language, if such an assessment even makes sense. I'd love to see a very careful discussion of this claim. 10/
As I said, a thought-provoking article, which is the best kind! Thanks, @GaryMarcus ! Looking forward to your book.
