Assholes like Musk will never be willing to pay more taxes for something like this. People like him can never get enough. Hence he and others try to keep people quiet with "you will be cared for" until it becomes impossible to hide that this is a lie, but it will be too late to really do anything.

"Smart" != "moral". Some of the most evil fucks have been pretty smart. Obviously, to be a smart evil fuck, you have to ignore some things, but high intelligence does not come with the skill to apply it competently and equally to all questions that arise.

Not true at all. LLM-type AI is fare more limited as even the dumbest fuck. The only thing it has is a stellar knowledge base. But actually doing things? Not so good. And understanding things? Zilch.

All the people fawning over LLMs do not really understand them or what they can do. Essentially it is somewhat better search with an NLP interface. Nice, but not a game-changer.

Indeed. As a group, humans are incapable to sanely evaluate individuals. The ones with the most money and power are almost always the scammers and the scum.

That crap is not even worth an answer. But thanks for identifying yourself as a clueless true believer.

Pick any two. I guess we know what choice Google made.

"Business"-oriented people hire not by competence but by fame. One reason they are primarily a problem.

Good point. Obviously, a counter example is very simple in "proof structure", especially as it does not need to tell you anything about what an optimal result would look like.

I guess Mathematicians will continue to have good job opportunities after all.

As to what that LLM does, there are a lot of non-statistical tools it could be using. Obviously, if they tell us, say, "the LLM handed 100'000 possible counterexamples to Wolfram Alpha and Wolfram Alpha picked the single one that was not nonsense", that kind of would destroy the picture they are trying to paint of their product. On a related topic, I am beginning to suspect that Claude Mythic may be running very conventional tools to help it find bugs in code.

That seems to be a newer thing. I had some other problems maybe 1-2 years ago and Brave fixed them. I do not mind using different browsers for some specific sites. Still, good to know, thanks.

Either that verification will take a few years or this was exceptionally easy to prove once the pre-requisites were clear. Hence at this time there are only two options: The proof is wrong or it was not hard to find for a machine that cannot reason, but can trawl though vast amounts of data looking for correlations.

My guess is the second: The only reason nobody else found this is because it is very easy to do, but the prerequisites are very non-intuitive and hence nobody looked. Kind of like those "20 year old" security bugs LLMs find these days. Obviously, the LLM scammers try to suggest that humans looked and failed for 20 years, when in actual reality nobody looked because it did not make sense to look. Occasionally, plausibility prevents smart humans from finding rather easy to find things hidden in a mountain of data. This is the one thing LLMs are somewhat good at: Better search. But they cannot reason or construct proofs from scratch or anything like it, even when that is implied in the OpenaAI press release. They can only find statistical relationships in data. That is useful, but but nowhere as groundbreaking as it gets presented at. And the applications for finding "new" stuff this way are very limited.

When Trump does it, it can obviously not be communism! Or at least that is what his mindless followers deeply believe.

Essentially, yes. The problem with "statistical inference" is that is does not stack. Logical inference can be stacked as high as you want and the end-result (unless you made a real "hard" error in applying the rules) will always be valid. With "statistical inference", that is not true. Each step only has a probability smaller than 1 to succeed and that is fundamental. At some, not very high, number of steps, the results become arbitrary. Hence the "feat" here is that the LLM found the result to be very close in reach with regards to inference steps.

So, why did no Mathematician find this? Simple: Mathematicians are smart and look at things that are intuitively promising, given a real understanding of the overall picture. In rare cases (and we have one here) that results in cutting of a short, but not intuitive and hard to see inference path. That is also the reason some coding models find very old bugs: These bugs were hard to spot for humans with actual general intelligence and insight, but relatively easy for a strongly limited inference mechanism that has no understanding at all and just tries simple (but non-intuitive) things in order of probability until it succeeds.

Example: The recently hyped FF bugs found by Claude Mythic were two use-after-free (and one more, 2 of 3 only "medium level, and about 20 others found by humans at the same time, which gives you an idea how much Claude covers in relation to human review) and these can be found completely automatic with, e.g., gcc -f analyzer or other tools. Not impressive, but apparently nobody used quite conventional tools or real code review to find these and they are _easy_ to spot. Other example: CURL, where Claude found one security bug (and 4 false positives) in a code-base that realistically will still have more like 100 left (which given the code size would be pretty good), but all not obvious to a human that looks for things that make sense and most (excluding that singular one) not within reach for an LLM either.

The whole thing, including this story here, is full-on lying by misdirection to keep the investor money flowing. Unfortunately, most people are not smart enough to understand what is going on and how utterly limited LLMs are. Well, _those_ people may be within reach of being replaces by an LLM, at least partially.

Vivaldi continues to mostly just work fine on most sites. Sadly, Slashdot is not one of them, here I use the Brave Browser for better experience.

You sound really stupid and disconnected here. I am neither "communist" nor "big government" and that is blatantly obvious from what I write. I am anti-stupid though, and that directly concerns you.

Obviously, the only punishment for insecure FOSS is irrelevancy.

That is true. But that is not as what this gets pushed and such a search tool would never even remotely justify the extreme effort LLMs need to do this.