AGI is never happening, is it?No matter how much compute they throw at the problem, there is always another trivial task they are spectacularly retarded at
https://x.com/alek_safar/status/1964383077792141390
>>106518499I dunno, if you told anyone 5 years ago that AI would take most artists jobs they’d think you’re crazy but now gen AI can produce in a minute a piece that is better than what 95% of artists can manage. The progress is clearly there even if there is some stuff it’s still bad at.
>>106518600That's why I said AGI. So far a general intelligence is extremely elusive. Any of these models embodied in the physical world would fail at so many tasks we find trivial.
>>1065186005 years ago i was wearing diapers
>>106518626It’s not that far off. Logic is embedded in language. There is no real ”thinking” only Logos and its expression
wasn't there something about how marketing material for watches always has the hands at certain times for aesthetics, so whenever a model sees a clock face its overfitted to those times
>>106518706ah interesting, boom theres the logical reason why they perform bad.OP and twitter vanity project guy is a faggot
With Machine Learning, absolutely not.It's fundamentally wrong, and as contradictory to how humans learn as you can get.The brute-force method of an AGI kernel will be just 1:1 cloning the smallest brain we can and scaling ASIC hardware to map larger and larger brains.The future-thinking AGI kernel will made be the one who looks at Nikolai Bernstein's research and realizes humans inherently encode and decode Fourier Transforms.
