Re:John Carmack's new AGI company, Keen Technologies, has ra...
by Jack000 on Friday August 19, 2022 @07:28PM
(#32527348)
Biology is not necessarily the only path to intelligence. Modern ML has diverged very far from biomimetic approaches at this point.
The brain does have more "parameters" than ML models but this applies just as much to humans as it does to animals that we don't consider particularly intelligent.
Re:John Carmack's new AGI company, Keen Technologies, has ra...
by jmyeet on Friday August 19, 2022 @07:36PM
(#32527417)
> Biology is not necessarily the only path to intelligence.
True.
> Modern ML has diverged very far from biomimetic approaches at this point.
I agree but my take on this is different: it just lays bare how simplistic modern "ML" is. And I put "ML" in quotations because most "ML" is really just statistics.
If you accept the premise that ML is a highly simplified version of the one model of sentience we have that works (and you might disagree with that) then that just puts us even further from AGI. Why? The information/entropy in the system.
Re:John Carmack's new AGI company, Keen Technologies, has ra...
by Tenoke on Saturday August 20, 2022 @06:44AM
(#32530959)
That's a poor example. We have plenty of models that vastly outperform OpenWorm on whatever task you choose. Their failure simply suggests AGI is unlikely to come from direct emulation of existing intelligence.
My evidence? OpenWorm [1]. OpenWorm is an effort to model the behaviour of a worm that has 302 mapped neurons. 302. Efforts so far have fallen way short of the mark.
How many neurons does a human brain have? 86 billion (according to Google).
I've seen other estimates that the computational power of the brain is roughly estimated as 10^15 operations per second. I suspect that's on the low end. We can't even really get that level of computation in one place for practical reasons (ie interconnects).
Neural structure changes. The neurons themselves change internally.
I still think AGI is very far off.