Gödel's Incompleteness Theorems vs. AI Minds

Watch the full episode with Dr. Mark Colyvan here: • The Most Fascinating P...

watch the full episode here: kzbin.info/www/bejne/gaK8mWSeZ9lrjbc

[@jimmymcjim1984]

Wouldn't it be that at any and every stage of the nested self referential computations that you propose, consciousness can "look" at the paradox/glitch from the outside (out to infinity). So consciousness in that sense is not computation. Keanu Reeves did a video on this. Roger Penrose speaks well on this subject.

[@iamjuleshimself]

Isn’t it great how you have almost 100 videos here discussing the good stuff with only 10k subs, and then you get 10 times more subs on your other channel just talking about notebooks lol

[@bookishbrendan8875]

Would love to see you dialogue with Pat Flynn over at Philosophy for the People, Parker. 👍

Then you'll love this 😄 kzbin.info/www/bejne/h3nInX-MidqJqNEsi=NlKjoH1DZ8dPumTr

[@LEMAN-AND]

Greetings, I use the same microphone as you. You have a very good sound! could you please tell me what settings you have? this will help me do my job better)

[@S.G.Wallner]

Great question. I hope the answer is...

[@QuicksilverSG]

The significance of Gödel's Incompleteness Theorems is not the conclusion you draw from the proof. It's that flash of insight you get after reasoning through the proof till you understand what makes it true. That's the point when you experience a thought process that cannot be algorithmically replicated by a Turning Machine.

[@aqua200546]

Thank you, thank you, thank you for this clip. The leap from Gödel's Incompleteness theorems to the conclusion that the human mind is non-computable is really seductive, but I think it's baseless. An average human can't look at a formal system and find prove or disprove its completeness or consistency in a flash of insight. This takes training, practice, and trial and error, and relies on learned heuristics, just like any human skill. But machines can also perform heuristics, and GPT's can learn new heuristics just by dialogue with humans (this is insanely inefficient but it does work). What evidence do we have that, say, GPT6 couldn't mimic the mathematicians reasoning process? At that point, what evidence do Penrose et al. have left to assert the non-computability of the human mind?