I just don't think a general intelligence is possible based on the current technology. What we call "AI" right now is just predictive text on a massive scale, there's no actual thinking going on. There would need to be an entirely new paradigm.
Comments
Log in with your Bluesky account to leave a comment
My point was more that like, I don't think General Intelligence is achievable with the current approach. I think it's going to take an entirely different technology to create a truly artificial intelligence. I did see that a company is looking into using organic materials, that might be something.
Ah, I think I took you too literally (esp. "lifetime").
I agree with you, yeah. Current approaches can't scale up to AGI level. Minor improvements to LLMs + more data only go so far. (And they're out of data.)
And like I said, it's kinda pointless. Why would anyone want an AGI over a non-G AI?
Lol yeah, I only said in my lifetime because I don't think they're ever going to do it with LLMs, but if they do it after I'm dead how am I going to eat the hat or whatever.
Comments
For comparison:
My dad was born when computers with 32KB of memory cost $8K.
His dad was already 10 when the first Bombe was built, and he died after GPT-3.5 released.
I agree with you, yeah. Current approaches can't scale up to AGI level. Minor improvements to LLMs + more data only go so far. (And they're out of data.)
And like I said, it's kinda pointless. Why would anyone want an AGI over a non-G AI?