487. Challenging AI’s Capabilities with Gary Marcus

In the last five years, artificial intelligence has exploded but there are a lot of holes in how it works, what it is and is not capable of, and what a realistic future of AI looks like.
Gary Marcus is an emeritus professor of psychology and neural science at NYU and an expert in AI. His books like Taming Silicon Valley: How We Can Ensure That AI Works for Us and Rebooting AI: Building Artificial Intelligence We Can Trust explore the limitations and challenges of contemporary AI.
Gary and Greg discuss the misconceptions about AI’s current capabilities and the “gullibility gap” where people overestimate AI's abilities, the societal impacts of AI including misinformation and discrimination, and why AI might need regulatory oversight akin to the FDA.
unSILOed Podcast is produced by University FM.

[@voyagersa22]

So in the end who is right about language?? Chomsky or Hinton ?

[@johan.j.bergman]

Chomsky is certainly more right, that's for sure. The pace at which humans learn language is just not possible without an innate structure for it. Where it gets complicated is to which degree. But it's not just operant conditioning, reinforcement learning. There's just not enough stimulus for that to be logically possible.

[@marbin1069]

Hinton, of course.

[@snarkyboojum]

34:25. Gary misspoke here. AlohaFold doesn’t work with nucleotides and “figure out their shape”. AlphaFold works out the shape of proteins from their constituent amino acids.
While he misspoke here, it shows a lack of general scientific understanding. That concerns me. If he confabulates these sorts of things so readily and confidently, what else is he getting wrong….

[@marbin1069]

This guy is not an AI expert.