If a massive vacuum cleaner absorbs the world's knowledge from the Internet and regurgitates it using the same AI technology we've had for decades; that's a level of theft the world cannot allow.
I have not read the essay, yet but when 'we' talk about
> reasoning errors,
we do not mean reason in some natural, universal, scientific kind of sense, right?
Given that the training data can only contain human reasoning and computational logic, reason in the sense of LLM's can only be interpreted as "rational facts AND nonsense humans made up to create systems that would support consumerism-driven sanity", correct?????
Please understand, I'm not mocking, I'm genuinely interested
in the ways human reasoning radiates into the code LLM's learn while they realize (the computational equivalent of a new-born's eyes opening) their cognitive (&) sensory
(that which triggers/causes/elicits/prompts/influences) their origins (every whatever-second/moment of their existence).
> we do not mean reason in some natural, universal, scientific kind of sense
I believe there are two different ways people think about this:
1) Some see "reason", "intelligence", "free will" and/or "consciousness" as emergent phenomena that arises naturally from normal physical processes (or they dismiss the concepts completely as illusions for the same reasons).
2) Other seem to consider these somehow independent from physics, or if not will tend to hypothesize that it is linked through quantum mechanics to something more fundamental.
If interpretation 1) is correct, then we will probably see full AGI in our lifetime. If 2) is correct, it could be that we can never create "real" AGI, or at least not without quantum computers.
I've never seen anyone in camp 2 come up with convincing definitions of the terms, though, beyond "I know it when I feel it".
Anyway, it's really hard to have a discussion with someone with the opposite conviction, since these beliefs tend to be held axiomatically and/or religiously.
It's a hype train (am I using the term correctly here?), a trend.
Would be nice to trace it back to it's origin. This Hasan-Abi Streamer comes to mind. I only watched like 2 x 5 minutes on two different occasions, on one of which he talked about 9/11 [1], the potential conspiracies of which were a long debated topic back in my Uni days. In those days I learned a bit about the history of the Israel-Palestine conflict and since then, for over a decade, watched the media and politicians distorting the history AND the narratives quite a bit. It's easy to create a trend here and it feels very much like the goal is again, polarization.
Was the massive wave of support for Ukraine a hype train? I mean, I guess so but my point is that it still has nothing to do with some sort of conspiracy theory. Maybe there's no narrative distortion, people really just do support them?
punishing "your own kind" and deriving pleasure from the suffering of all, including yourself, because you feel like you all deserve it.
long episodes of constantly recurring pain and an increasing intensity of even small doses of pain acting as a signal that you did wrong and should feel the consequences.
We expect him to lie whenever the board thinks it's necessary and we expect him to tell the truth whenever it fits the narrative.
We also expect him to play along, even when some feature is too freaking powerful or so fucking pathetic it would only make marketing people and influencers drop their panties and write 15.000 fucking newsletters about it because PR.
The company is about money and he simply didn't prioritize that. He tried to blow it up, exalted, exaggerated, trying to make people aware of the fact that OpenAI has no edge on the competition at all.
There are so many options and OpenAI got waaaaaaay too much attention.
use research and AI to analyze Sutskever's character. the way he talks, the way he writes, what he did in the past, where he studied, who he was and is "acquainted with" ... do the same with the rest of the board and with Altman as well.
someone hire some PIs so we can get a clear and full picture, please & thank you
