You can buy blue checks, I guess. On the other hand they shut off embeds and access to replies unless you were signed in so it's functionally dead as a "website". Oh and sometimes there's child porn? So I guess it was overkill unless you care about things like moderation and safety. Anyway, excited to see how it very fairly handles the next US elections! I'm sure most of the remaining devs have invested their time there.
Not really the point. I think Musk just wanted to trim headcount down to something that could keep the product running, more or less, and get rid of all the costs he could. He didn't care about turning Twitter into a hugely successful business or an amazing product. He just wanted to be able to control and influence what people say on the platform, and push his agenda and politics.
It's how normal people use the word: doing something specifically to inflict pain on a human or animal. Slaughtering animals for meat isn't torture. Keeping them alive while inflicting pain because you enjoy the experience, is.
Have there really been no other more interesting war games in the last quarter-century, or did all the negative attention this got just result in us never hearing about another one?
There have been far more interesting war games in the last quarter century.[0]
This one shows the the US narrowly winning against China in a conflict over Taiwan. The US wins but with tremendous losses -- specifically in the form many munitions that take years to decades to replace.
And it just so happens that we witnessed a conflict play out just a few months ago and that resulted in a similar depletion of munitions albeit with minimal losses of American ships and aircraft.
What's very troubling about this is that in response the US moved munitions from the pacific into the middle east, leaving Taiwan, South Korea, and Japan in a very vulnerable position.
This may explain why the current US president was unusually obsequious to the leader of China when in the past he had been particularly bellicose.
Also, cool fact: While researching this subject I learned that the engines for most American cruise missiles come from a single company.[1]
I've heard it said that such systems may be used by militaries, where they have an organizational structure naturally-suited to large keystream distribution.
Unlike e-commerce, it's no problem to physically send the proverbial officer handcuffed to a briefcase to the nuclear submarine before it submerges for 6 months.
Also the messages to be secured are, um, short and... infrequent.
"Contrary to how it may seem when we observe its output, an LM is a system for haphazardly stitching together sequences of linguistic forms it has observed in its vast training data, according to probabilistic information about how they combine, but without any reference to meaning: a stochastic parrot."
So perhaps this has always been a negative claim, about what language model AI is not.
> "Contrary to how it may seem when we observe its output, an LM is a system for haphazardly stitching together sequences of linguistic forms it has observed in its vast training data, according to probabilistic information about how they combine, but without any reference to meaning: a stochastic parrot."
and
> "Meanwhile you have multiple Fields Medalists (Tau, Gowers) saying they’re very impressed by LLMs’ mathematical reasoning, something that the stochastic parrots thesis (if it has any empirically-predictive content at all) would predict was impossible. I doubt Tau and Gowers thought much of LLMs a few years ago either. But they changed their minds. Who do you want to listen to?"
I don't understand how these things are supposedly incompatible.
Larger models and further other refinement reduce the "haphazardness" of produced text. A big enough model with enough semantic connections between different words/phrasings/etc plus enough logical connections of how cause and effect, question and answer, works in human language can obviously stitch together novel sequences when presented with novel prompts. (The output was not limited to sequences of n words that appeared 1:1 in the training data for any n for at least three and a half years now, if not even back to when the paper was written.)
"without any reference to meaning" veers into the philosophical (see how much "intent" is brought up in the linked post today). But has anything been proven wrong about the idea that the text prediction is based on probabilistic evaluation based on a model's training data? E.g. how can you prove "reasoning" vs "stochastic simulated reasoning" here?
Perhaps a useful counterfactual (but hopelessly-expensive/possibly-infeasible) would be to see if you could program a completely irrational LLM. Would such a model be able to "reason" it's way into realizing its entire training model was based on fallacies and intentionally-misleading statements and connections, or would it produce consistent-with-its-training-but-logically-wrong rebuttals to attempts to "teach" it the truth?
Maybe, but a claim about what and LLM is not is still a claim about what it can or cannot do. And specifically:
> without any reference to meaning
is vague, but I read it as actually quite a strong claim about the limitations of LLMs. I don’t think it would be possible for LLMs to do long chains of correct mathematical reasoning about novel problems that they haven’t seen before “without any reference to meaning.” That simply isn’t possible just by regurgitating and remixing random chunks of training data. Therefore I consider the stochastic parrots picture of LLMs to be wrong.
It might have been an accurate picture in 2020. It is not an accurate picture now. What is often missed in these discussions is that LLM training now looks totally different than it did a couple years ago. RLVR completely changed the game, allowing LLMs to actually do math and code well, among other things.
reply