[neffy]

No, it´s really not - it's exactly what they are. Multi-dimensional pattern matching machines, using massive databases put together from resources like stack overflow, Clegg's (every cheaters go to for assignment answers, massive copyright theft etc.). If that wasn´t the case, there wouldn't be jobs right now writing answers to feed into the databases.

And that´s actually quite useful - given that most of this material is paywalled or blocked from search engines. It´s less useful when you look at code examples that mix different versions of python, and have comments referring to figures on the previous page. I´m afraid it becomes very obvious when you look under the hood at the training sets themselves, just how this is all being achieved.

[atleastoptimal]

Look into every human’s brain and you’d see the same thing. How many humans can come up with novel, useful patents? How many novel useful patents themselves are just variations of existing tech?

All intelligence is pattern matching, just at different scales. AI is doing the same thing human brains do.

[omnicognate]

> Look into every human’s brain and you’d see the same thing.

Hard not to respond to that sarcastically. If you take the time to learn anything about neuroscience you'll realise what a profoundly ignorant statement it is.

[weweersdfsd]

If that is the case, where are the LLM-controlled robots where LLM is simply given access to bunch of sensors and servos, and learns to control them on its own? And why are jailbreaks a thing?

[ethanwillis]

Seeing as your LLMs need the novel output of human brains to even exist or expand capabilities, quite a lot.

But even if it's not a lot, it's more than the number of LLMs that can invent new meaning which is a grand total of 0.