Too late! I already watched it (in 1902).

Some important background for this is that, around 1900, mathematics got formalized to the point that there is a standard for whether a proof is valid or not that doesn't depend on humans thinking about it, just doing a bit calculation following an algorithm. However, it is such a large calculation that nobody wrote complete formal proofs or verified all of the steps (since they'd be spending huge amounts of time on obvious parts, and not the novel aspects). Then, in the past 20 years, people have made computer systems that work on complete formal proofs while showing you the interesting parts, with the rest in macros, and check these proofs automatically. This means that finding proofs has the form of things like what AlphaZero is good at: something to check if moves are legal and a win condition. However, only a relatively small portion of mathematical research has been put into Lean so far, and that doesn't include the sorts of obscure results that often turn out to be useful in other problems. So the current wave is using LLMs to find results in the literature that might be relevant and producing formal versions of them, game AI to put them together, and deterministic validation to check each of these as it goes along and then people can see at the end that the calculation shows the proof is valid. The only part requiring mathematicians is checking that what the proof proves is what they understood the open question to be.

Thanks for sharing!

Ironically, the industry trend of doing SaaS has pushed the need for customer support way up, compared with the traditional model of selling a finished product once and issuing refunds or replacements in rare cases.

That's what they did the last time. And here we are again. And there we'll be again in the future.

None of that responds to my point.

Firstly, US courts have no jurisdiction in most places, so what.

Twoly, copyright is governed by international treaties which *are* ultra-restrictive by their nature. Wishing it wasn't doesn't change the fact that stuff written online has automatic ownership unless explicitly unrestricted. And no, it doesn't need to be strengthened, it already is strong enough. Some say too strong.

Threely, not looking to get rid of AI, looking to make the pirate companies pay, because...The Law. If they disappear, they can be replaced by a new generation of law-abiding AI companies.

Fourly, open source code is explicitly unrestricting copyrights under specific licenses. But open source AI is no such thing, as it's initially copyright encumbered too, due to training issues.

Fifthly, do I personally care about open source AI? No. I care about scientific papers, and I care about new ideas, but I view current state of the art AI as statistical toys. I'd actually like to see *real* progress, which isn't possible as long as everyone rehashes the old ideas (with more GPUs and more bandaids).

Anyway, that's what *bad* AI means to me. That was the topic, no?

The Indian AI call centers didn't last, people didn't like them. The Artificial ones won't last either, for the same reason. Everything old is new again.

The real world locations for shooting are often a result of financial incentives. The studio gets some funding from the local government responsible for the shooting location. It's no different from paid product placement by large private companies, except the product here is the surroundings and the play is for tourism dollars.

Sloppy AI generated cities would be pointless, and quite possibly misrepresenting stakeholders.

Bad AI includes AI that infringes copyrights during the training phase. If this AI has been trained (even partially) on content found on the Internet, then it's bad AI. (*)

(*) There is no fair use or implied consent for documents sitting on web servers around the world. Either there's an explicit licensing agreement by the actual copyright holder that permits you to copy the file to your computer, or you're infringing, Mr AI.

Don't forget that training is a one-off snapshot in time. That's why training must be repeated regularly. There's no learning in between trainings for foundation models.

generally-better-than-a-large-pc-of-the-population does not mean good-salary. For a good salary you'd have to be in the minority at the top, where statistics breaks down.

Hear, hear! I was able to poop this morning too, it was wonderful... Kids these days just don't know the value of things!

"You" are already committing money to them, through "your" pension funds, via SpaceX, and other vulnerable F500s. The game on "you" has already been played, because xAI is how SpaceX is connected through the maze of fake IOUs.

The underlying isomorphism point is solid imho. And I'm not actually suggesting that a "team" of AI coding agents is a thing that works, just that agentic AI software development has the same flaws as human software development.

The pattern of use by software engineers, who are by and large unqualified to lead teams of humans, is to use teams of agents any way they can, currently (observations in an industry with unlimited money resources).

There's no rhyme or reason to either their own management of the AIs for their projects or the industry's own attempts to delegate sub-agents automatically. It's purely "throw AIs at the problem" and "throw more AIs at the problem to fix the AI created problems". Exactly what MMM warns against.

I take back my mention of answers. It was meant to be vaguely sarcastic, and you were right to call me out. I should re-read MMM.

Writing code is not how or why the bubble would be crashing. Not making enough money for impatient investors is why the bubble will crash.