3 ms·
1) It may be possible to brute-force a model into something that sufficiently resembles AGI for most use-cases (at least well enough to merit concern about who
by captainclam 3y ago
1) It may be possible to brute-force a model into something that sufficiently resembles AGI for most use-cases (at least well enough to merit concern about who controls it)
2) Deep learning has never been terribly scientific, but here we are.
- cactusplant7374 3y agoIf it can’t digest a math textbook and do equations, how would AGI be accomplished? So many problems are advanced mathematics.
- captainclam 3y agoRight, I do agree that the current LLM paradigm probably won't achieve true AGI; but I think that the current trajectory could produce a powerful enough generalist agent model to seriously put AI ethics to task at pretty much every angle.