12 comments

[ 22.1 ms ] story [ 2326 ms ] thread
It's inference. It's really good at generating stuff when the example base is extensive. Like for non-esoteric coding.
Is a brain also inference? I know that an LLM works very different from the brain, but I wonder what makes a brain more capable of thinking. Is it the long term context? Is it a different type of neuron activation?
Ha before reading the article I thought "this must be an interview of Lecun". A bitter scientist that hates he was left behind the revolution.
The article seems to define "smart" as being good at spatial awareness and navigating a body through 3D space and such. Thus, a mice is smarter than an LLM.

That's the first time in my life I hear this definition. Until now, the word "smart" has meant doing exactly the things LLMs do, and mice don't.

I guess it is a sign we are re-evaluating what makes humans special.

i guess inference engineering, like dpsark or dflash specific speculative decoding technqiues
Yann LeCun was saying 3 years ago that because token generation is auto-regressive, its mathematically impossible to generate a long stream of coherent tokens, because errors amplify exponentially.

and then models learned that they can back track and error correct

so much for "mathematically impossible..."

It depends on how you define "smart".

For me, "smart" means doing things less based on instinct. Things humans can do but mice cannot, things mathematicians can do but normal people cannot, etc.

Considering the unit distance conjecture was disproved by OAI's model last month, I think maybe LLMs should count as "smart".

Besides "smart", the headline also conflates AI with LLMs. The real, non-clickbait title is "Yann LeCun, founder of AMI Labs, is developing a new AI system"
Its definitely not as dumb as MAGA crowd
What's next is more AI spam-slop. I already noticed this on youtube. Tons of short videos are AI sloppified, making youtube worse in the process.
So we have that quote from the Oxford guy about explanations: "systems that can explain... You need models that can answer questions like: What matters? What causes what?", and then a mention of simulation of what the world looks like.

Fine, that describes theorizing.

But then a contradictory ending statement: "We're still going to need humans to figure out what questions to ask, what to build, what to create".

So that's moral theorizing. I don't think you can have one without the other. Then there's two more suggestions before the end of the article:

> smarter than us

> staff of assistants

Both of which are completely gratuitous assumptions. Why should its theories be better than established ones? Is it supposed to be a maverick hermit genius and come up with everything from first principles, or does it in fact participate in the existing world of ideas like a normal person? Then, being a normal person with moral theories, why would it take on the role of assistant rather than theorizing "I don't want to do that for you"?