Hinton vs LeCun vs Ng vs Tegmark vs O
Hinton vs LeCun vs Ng vs Tegmark vs O
Three top ML researchers, a leading physicist, and a former French Minister, at each other’s throats
If last weekend’s OpenAI drama was top-billing, the can’t-look-away undercard was four heavyweights, Hinton, LeCun, Ng, and Tegmark slugging it out on X, briefly joined by former French Minister Cedric O, now working at a French LLM startup, Mistral, and arguably having pulled a 180 with respect to regulation.
The main issues were two: does deep learning understand anything, and how should we regulate AI. Some spicy samples:
Andrew Ng weighed in, too.
And so did DeepMind exec Nando de Freitas, rather soberly:
Menawhile Tegmark and LeCun had some back and forth
which eventually led to the foreign minister rushing in, snidely
and LeCun rushing to his defense.
And that’s just some of the highlights.
§
There were no winners here. LeCun showed his customary lack of humility, and a lot of people pounded on Hinton even though (in my humble opinion) what he said about LeCun is true, inasmuch as LeCun doesn’t strike me as a particularly great listener.
Meanwhile, Hinton doesn’t have a real argument on understanding, just an intuition. I hate to give a round to LeCun, after all we’ve been through, but he’s right on that count. (On regulation and AI risk, he’s the one without a plausible argument, just a hope that everything turn out alright.)
Bing,powered by Dall-E 3, was an even bigger loser, drawing “Hinton vs LeCun vs Ng vs Tegmark, boxing” with four boxers not boxing, not one of them looking remotely like any of our heroes:
§
Actually, there was one clear overall winner: Tegmark. As much as Ernie Davis and I utterly disagreed with Tegmark’s more technical map of the world/LLM world models paper a few weeks ago, pretty much every word he said this week about regulation was true.
And this is the big week, do or die in the EU. Either we walk away with some real regulation on generative AI, or we don’t. And I do mean “we”; what the EU decides will affect all of us.
If you know anyone in the EU government, now is the time to let them know that you don’t want big tech to run the world.
Gary Marcus is amused to see LeCun taking a Marcusian position on understanding, and thrilled to see Tegmark is beating up on LeCun where it counts.
During an interview with NDTV last week, I said that if in 1950 you polled 1000 physicists globally, you would find universal agreement that nuclear weapons posed an existential threat. (Though some would also undoubtedly say they were a necessary evil, for deterring future war).
Yet I then said that if you were to poll 10 AI luminaries today, you will get at least four different opinions on whether AI poses an existential risk.
That’s not a problem necessarily. It’s the nature of the AI beast today. But it’s hard to criticize Congress (and other global policy bodies) for either under- or over-regulation of AI when there are widely different opinions on such a fundamental question.
Lol! Who needs AI to doom us when we have jokers like this lot pulling the strings?