Although saying that, even the "stuff I did in the 80s" parts show an overly mechanical misunderstanding of how the human mind works (not to mention his weird take on what constitutes an emotion), so maybe a lot of these "pioneers" have been wrong all along...?
When you've already decided you're playing with the building blocks of life itself, every wrong/incoherent/meaningless answer an algorithm retrieves and strings together can just be more evidence of it struggling to live
Out of that whole interview one paragraph makes sense and I totally agree with it. The rest is cuckoo for cocoa puffs. He seems to think human consciousness is no different than a back-propagation neutral net? I think? Like he thinks he discovered the secret to AGI in 1986 but no one understands it?
He does not understand any approach to communication other than semantic chaining to maximise the vibes.
That does not mean no other approach is possible.
my god it’s just absolute crap after absolute crap. how are the so-called “smartest” people also sometimes so unmoored from reality? I’m worried I’ll bump into him some day lol
I wanted off, 2-3 paragraphs in, the second he said a robotic arm expressed emotions by smashing a pile of parts because it's vision system couldn't make something out. The interviewer asks: was it programmed to do that?
Hinton: yes
🙄😤
Yeah but were they venerated as geniuses, with associated bad habits, before having a cognitive decline?
I've seen a video about former respected physicists becoming "cranks", there's a pipeline there.
and here we see geoffrey revealing himself to be just another mediocre white guy, whose mediocrity was celebrated only because it was allowed to flourish under conditions that excluded nearly everyone else who would have overshadowed him, reducing all of humanity down to the inhumane
typical, really
Ah right, Hinton doesn't appear to understand cognition or consciousness, in addition to AI research, public policy, and all the other things he demonstrably doesn't understand. How did he ever achieve the reputation he has?
Because when the world was new, you could get very far on coming up with super basic stuff.
See, referring to a different context, Mickens (2013)
www.usenix.org/system/files...
That is so much nonsense. Also, is it really a new thing from this month? "Letter last week" sounds like 2023, but the date is June 12, 2024, so I'm super confused.
And the story about seeing a robot have a feeling back in 1972 ... I wonder if he really thought so back then too (has always been this far down the rabbit hole) or if that's some revisionist history about his own thoughts.