@elbutterfield Alfred referenced Soul Machines as an example - they're amazing and pioneers of the hyper-realistic avatar. Here's an example:
@Zevon @elbutterfield at 0:43 they hit on the key missing ingredient IMHO* opinion. Without a body, any machine intelligence will be a shallow faint echo of what we recognize as "intelligent"
Very curious how they've implemented embodiment.
*could be humbler
**fricking self driving cars don't even know their own dimensions or weight gtfo
@b4cks4w @elbutterfield all fair points :)
here's how they've built their cognitive models--intense!
Functional Layers of a Digital Person - Soul Machines
@Zevon @elbutterfield they're coming from neurology, but how do they translate that to code. I'm unconvinced it's possible or even the right approach -- our brain architecture is completely different from binary computers. If I had the smarts and money I'd take a different route and play to the C/GPU strengths instead of trying to emulate a massively parallel network mess of a thing (brain)
lol but no one listens to me for good reasons.
@Zevon @elbutterfield Well my premise is: forget it, stop trying. Make an intelligent machine. That will seem as foreign to us as an alien lifeform...because it will be one.
And apparently nobody wants that 😂
IKR! I had this convo yesterday with a fren. It's like Musk, Altman, Hinton and their predecessors want to be gods who give birth to a life form. Like becuz they don't birth and nurse other humans (though they play a critical role) they chase this dream of replication @b4cks4w @elbutterfield
@Zevon @elbutterfield might just be tunnel vision of computer scientists (no that doesn't explain Musk but pffft, he's a poseur) but I'm perfectly comfortable with your analysis: men being men. DAIR is mostly (all?) women saying, no, actually...
https://www.dair-institute.org/about/
tx for mentioning DAIR 👍🏼 I've followed them for awhile. @b4cks4w @elbutterfield
@Zevon @elbutterfield just an old(er?) caucasian dude trying to expand my bubble. No tx needed.
lol that's the BIG proprietary conundrum. They've been working at it for over a decade with the current results. Your approach sounds interesting. @b4cks4w @elbutterfield