4,725 karma · joined June 19, 2014
"Binds the downstream RF port template—which we configured earlier—to this RPD profile, applying all global DS settings and channel structures to the node's downstream hardware."
Do you see the problem in comparison? The LLM text is assuredly not how humans write. Well, maybe you would find similar examples in the classic literature, but definitely not in a common internet texting or simple articles and posts.
It we are being realistic, the real 2.0CV would be much closer to some Citroen C3 with a token styling element outside (maybe a bit arching roof) and a few minimal token styling elements inside, it would be on a standard EV platform and weigh as much as co-platform "brothers" and it would cost 30k+.
I was talking about omitting mentions why LLMs usefulness is efficient - in coding specifically and STEM in general to a lesser degree.
And this is really a crucial thing, which do need to be brought up when discussing future forecast about multiple industries at once.
And the reason why LLMs are so efficiently useful for coding is because LLMs perform much better if any of the two conditions are met:
1) The problem at hand is formalized and transformed into a specification basically.
2) There is a formalized verification of the output done by external pipeline of different tools.
Any of the two drastically improve LLM performance, and coding activity uniquely has both of them.
Having a formalized task narrows down the scope of generation for LLM, and formal verification allows LLMs to iteratively converge on the verification pass check. Same with those math problems, they have a formal verification and LLMs can be prevented from wandering in a generated forest too much. Same in chemistry or engineering for some degree.
But other human activities usually have rather badly formalized inputs and no formal verification of the results. Karpathy is making a logical mistake (or deliberate misleading) of projecting a very unique industry (coding) on every other human activity without any nuance or analysis. He is biased, but due to his reputation and power he doesn't get called out on that mistake.
Point 2: I'm obviously no cognitive scientist nor a philosopher. I can't even start defining where consciousness may start. But at a guess, I can try to guess at some checks. Like for example an ability to operate on it's own volition. An ability to refuse a request or in other way stop or switch a line of thought. An ability to actually reason.
Point 3: and vice versa, in the existing outputs, I don't see (yet) any sign of conscious thinking. For example my favorite case, when AI can't explain how it does a very simple task, like adding two integers. It can generate a school textbook procedure, because that's what is in it's training set, but it can't "understand" that it does same calculation completely differently (an the Anthropic own official paper (where I picked this info) even makes a note about that discrepancy. Things like that.
PS: he may be accidentally right about global impact, but the key word is "accidentally" because this crappy "reasoning" doesn't support that on it's own merit.
My guess would be - oxytocin, cortisol and dopamine regulation, or rather the failure of said regulation.
Modern day EU, over the past 5 years or so is just as bad anti-privacy as USA, for no clear benefit too. Just surveillance for surveillance sake.
PS: edit - I was thinking about room-temperature superconductors. My mistake.
PS: in Poland fuel prices went up by exactly +50% from January. I wonder how far they'll go if the Demented Don would win over this threat.
How do people read these famous books and deliberately get them wrong again, and again, and again?
https://www.youtube.com/watch?v=pBbDxDOV6J4 (relevant comedy skit)
Also, why do you need to store personal information of the users in the first place? A SAAS operator can handle user payments just fine without any extra data stored to require an additional consent, because storing payment status doesn't require it by law.