Alright I guess

I wouldn't put too much stock in many consumers' delineation between companies either. Anthropic markets differently than OpenAI, and so forth, but the general vibe really is the same: things will just flow between you and Claude/GPT/Gemini.

More to his and your pints though, the stuff around the "edges" is still interesting, potentially useful in its own right. But ultimately there's never a knockdown argument for or against any proposal, which leads me to be *extremely* cautious of claims made from on high wrt AI.

My honest response to all this is: I just don't know what to think.

I might feel differently after properly reading Block's piece, but the most we seem able to do with consciousness is work around the edges; never really getting a better grip on what it's supposed to be.

I've had mostly vague suspicions about the (seemingly widely accepted) idea that consciousness can arise, in principle, on any physical substrate.

I doubt there's a way to prove it can't arise on non-biological substrates, but the in-principle argument gets over-extended IMO.

Very disappointing. But it is sort of funny that Chomsky doesn't miss a chance to dunk on his critics (right after Epstein offers him the use of his NY apt? hard to parse)

Reading this and I'm remembering Summers whining about the Biden admin not listening to his advice in 2021-22.

The Origins Project exists FWIW, at ASU. 2010s sometime. Easily found, but they removed them from the official channel because of Krauss' stuff.

And his response was basically something like: 'I don't have to tell you about this stuff because it's my business.'

And yeah, moving your late wife's money is your business, but did that money really have to be moved by Jeffrey Epstein, Noam?

Chomsky never had accusations against him AFAIK. His Epstein connection (financial) has been known for a while.

www.wsj.com/us-news/jeff...

("relevant to us" only meaning it's comparable to the biological/human intelligence we are most familiar with.)

But all I'm really saying is, the paradox is useful only insofar as we apply it appropriately. I'm not sure that's been the case amid the "human cognition is actually really cheap to replicate" mania/depression/hubris of the past couple of years.

LLMs' essays are almost always impressive in the "it's amazing what they can produce" sense, rather than the "making conceptual progress" sense.

This can quickly turn into a can of worms that I don't want to open, because many would likely say they do this already (or that humans don't do this).

You often hear: it's easier to train AI to write an essay than do the dishes. The idea being that we mistakenly believed that writing an essay is uniquely difficult, and LLMs have shown this to be wrong.

That might well hold up. But a big reason to doubt it is that...

Someone's gotta go back and warn this guy about a visionary named Sam Altman

This also provides what I imagine is still a helpful taxonomy of Neuro-Symbolic AI, and probably relevant to claims that transformer-based LLMs are/are not Neuro-Symbolic.

More seriously, just imagine losing LeCun to Alexandr Wang. LeCun's got plenty of problems, but you're willing to spend billions on AI just to lose actual talent??

"LeCun’s next endeavour is focused on furthering his work on world models, according to two people familiar with the matter."