3 ms·
You're reading a whole lot into a tweet, in his interview with Dwarkesh Patel he says, about 20 different times, that scaling LLMs (as they are currently conce
by imperfect_light 2y ago
You're reading a whole lot into a tweet, in his interview with
Dwarkesh Patel he says, about 20 different times, that scaling LLMs (as they are currently conceived) won't lead to AGI.
- anoncareer0212 2y agoYou keep changing topics so I don't get it either, I can attest it's not a fringe view that the situation is interesting, seen it discussed several times today by unrelated people.
- imperfect_light 2y agoHe's said it pretty clearly, an LLM could be part of the solution in combination with program synthesis, but an LLM alone won't achieve AGI.