Jon Stokes
Jon Stokes

@jonst0kes

10 Tweets Dec 23, 2022
Again, for any of this to definitely be true, you need a model of "intelligent,"which Chollet does not have & which nobody has. How does he know that the brain is not "a record" to use his language from this thread.
He may be right, but he can't prove it. He also may be wrong.
On some number of levels, yes of course these models are not brains. But if you're going to make the strong case of "it's just math, unlike the conscious mind" then you really need a way to demonstrate that the conscious mind isn't the product of similarly structured "math."
I find this discussion as tiresome as others find "AI hype." If you don't know what the conscious mind is doing, then stop insisting that what an ML model is doing is totally different than what the conscious mind is doing.
That you shouldn't do this seems obvious to me.
When I write articles or tweets, I really, truly don't know where this stuff comes from. I don't know what I think until I start typing or talking. I spit this stuff out, then read it and go "yeah, yeah... I agree with this text!"
It's very much like using generative AI, frankly
I have pretty much zero access to my own creative process. I can sort of prime the pump a bit by going for a walk, or drifting into a twilight sleep state, or other rituals. But again, I don't know what's in my own mind until I dump it out into a text buffer or speak it.🤷‍♂️
Using generative AI tools is the first time I've done something that's not me being creative that is /like/ me being creative. So what I think about these tools is, maybe they can help me understand what's happening in my own brain. Maybe we're doing the same thing.
I've never had an external tool or system before that seemed to be creative in the way I am creative. So I find these models useful to think with (more here pairagraph.com), and if you're going to tell me why they're doing what I'm doing then plz tell me what I'm doing.
To be clear, I'm not naive about the math. I have a BSEE with a CS focus, and a minor in math (graph theory & combinatorics focus). I have done a lot of the relevant math, everything from matrix math to graph theory to filters & transforms. So this isn't math mysticism on my part
What if the chaotic soup of quantum reality (particle/antiparticle pairs popping into & out of existence etc) is noise that I'm running a bunch of diffusion steps on to structure into a sensory experience? Who knows, but "diffusion" offers me another way into this question.
Why is this not at least an interesting way to think about the problem of how we structure the world out of the apparent chaos of particles, waves, & other stuff? Are you sure our brains aren't doing that's meaningfully like this on some level? I don't think you are.

Loading suggestions...