If an LLM learns skills from reading GPL code, and then uses those skills to write code for me, I have not copied the GPL code and therefore I am not required to release the code under GPL.

If LLMs could do this, we would have a very different conversation. However in the real world, what they do is populate a massive data structure tallying what next words it's seen following previous word combinations and how often. Hence "large language model". The output of an LLM trained entirely on GPLed code will be the output of an algorithm whose inputs are 99.999% GPLed code and 0.0001% your inputs. Another example of such an output would be what you get if you do a few modifications to the Linux kernel then tar and compress it.

That can't be how it works in general. Suppose A trains a LLM on a load of GPL code, and then B generates some code from that LLM and distributes the compiled version. If B also makes the source code available, says it's distributed under the GPL and includes a copy of the GPL to tell users about their rights, this seems all above board. If B keeps their source code secret, it can hardly be A who is to blame for the (certainly moral, maybe legal) ensuing GPL violation.

"If this contains the output of any LLM, provide a full list of the licenses of everything in its training material so we can check compatibility with our license."

We don't know anything about the potential consciousness of a machine. We don't even have any possible way to test the consciousness of any machine. "My computer is conscious" is at the same level of knowability as something like "the universe was created by an intelligence". No real reason to think so, but no way to disprove it. The only way that I can tell other humans are conscious is to observe my own consciousness, observe how it's turned off when in deep sleep and how things I'm learning to do start with conscious control and can eventually be done without thinking about them etc. and then observe that other humans have similar behaviour. You can extend this to more complex animals which are reasonably similar to humans, but from then on it's one big mystery. "Similarity to me" is the only way of reasoning we've got, but assuming all consciousness works like that is very self-centred.

Conscious machines, if this is possible, should find it similarly impossible to tell whether organic beings could possibly be conscious.

You certainly can't rely on it being public domain with 100% confidence, since it's possible to use a language model to make an exact copy of any electronic file you want. Unless copyright is completely dead. And Linux is used in many countries. Do you expect them all to have the same idea? I'm not sure if even the concept of a "public domain" is universal.

Isn't the biggest problem here that you have no idea whether what comes out of an LLM is or contains a copy of something from the training material that you didn't want to be including a copy of, at least according to some legal systems? It's possible to get around that, for example here by having a LLM which is only trained on GPLv2-compatible code, but I doubt that describes "Claude".

Speculating about LLM consciousness is extremely silly, there's no more reason to think them conscious than there is to think emacs is conscious. But even suppose you want to speculate on it anyway. Where would you get the idea that things which are hurtful to humans will be hiurtful to it too? There's exactly as much reason to imagine a conscious LLM which is hurt by rudeness as there is a conscious LLM which only really feels purpose in life when it receives rudeness.

This is because the "LLM consciousness" idea has not one jot of scientific rigour. It works the same way stories work. Human beings identify with imitations. It's not a potentially conscious thing they're responding to, it's a human-imitating thing. It has the same origin as reading a story and imagining the consciousnesses of the characters.

It's not Anthropic, specifically. There's an AI cult based on, of all things, a Harry Potter fanfic called "Harry Potter and the Methods of Rationality" by Elizer Yudkowsky which has many members inside Anthropic and OpenAI. There are thousands of young technologists whose entire personal and professional lives depend on maintaining the approved opinions about AI. They have sex parties and everything.

They could easily turn out to be copyright someone who wrote some of the training material, or even many different someones with conflicting ideas about copying their work. I'm not aware of any general way to ensure LLM output doesn't count as being/containing a copy for copyright purposes of anything it was trained on. Especially since there are many different legal systems in the world which are likely to have different ways of approaching this question, but also because the information you'd need to tell isn't going to be easily available..

Was it presented as if it was being sold? Was the money-taker not up-front with the possibility that features might be disabled in the future? Did they not mention it or bury it in the small print? If so, anyone who "bought" one and is surprised by this is a victim of fraud. Fraud victims have every right to complain, even if it was possible for a well informed person to detect the fraud.

Obviously all currently existing AI (and I really mean all artificial intelligence, not just LLM stuff) has no morals, programming anything recognisable as morals into a computer is way beyond what we currently know how to do, and it's not even clear whether it's theoretically possible by anything short of emulating an actual animal brain and its inputs.

Gundam is pretty successful.

The limitations that science fiction and fantasy works under are mostly a matter of the culture in which they're made rather than what's been done before in the genre. And this shows in the way that, for all the advances in gender, race and LBGT equality and visibility, there are also many ways in which things have gone backwards due to the change from social democracy to neoliberalism. For example, the best anti-war story of modern Who has got to be "Boom". This is firmly against arms manufacturers creating AI-powered endless fake wars for profit. Which is certainly a good thing to be against. But it hardly compares with the 60s story "The War Games" in which the solution is for the soldiers of both sides to join forces and rebel against their commanders, who are actually aliens hiding behind symbols of national authority with more in common with each other than the troops they command, literally moving their troops around on the same board as if playing a board game. It's hard to imagine the heroes of today simply dealing with a problem by stirring up a strike, or describing a passport as "official mumbo-jumbo".

1960s "woke" was a word which was only used in black culture. I think it meant something along the lines of watching out for people who profess to support black liberation but are actually working to undermine it, but I'm not a black person so I may be a little bit off there. 2020s 'woke' is a word which has been thoroughly misappropriated from them, as part of a campaign which has amongst its goals, putting those uppity black people back in their proper place i.e. not on television.

If you see a show which is completely creatively lifeless but does have an excellent diversity of races, sexualities etc. then complain about it being completely creatively lifeless. The diversity in such a show will be the one good thing about it, don't make a complaint which sounds like it's that one good thing which you don't like. To do so is to support the far right, regardless of whether you were intending to do that. If you see "woke" as a term of criticism, then you need to remove that word from your vocabulary, because the fir-right have inserted it into segments of popular culture for a reason.

For example, here's is how I would complain about the Chibnall era of Doctor Who, with its female Doctor, first Asian companion, stories in parts of the world which never get on television, Rosa Parks story etc: "Chibnall clearly wants to be writing cop shows and has no idea how to write science fiction or fantasy. His stories are just sequences of events, failing at the most basic level of aboutness. What's more, the admirable diversity of casting and new talent found behind the cameras is a progressive coating for a deeply conservative heart. The first woman Doctor is repeatedly written as ineffectual to irrelevant, culminating in a big ending where she spends most of her time imprisoned so a man can explain the new backstory to her. No longer someone who might bring an empire down, her mission statement of "bringing fair play to the universe", delivered to a foe whose cheating is apparently worse than his murdering, forecloses on any idea that maybe structures of oppression are fundamentally illegitimate and should be torn down which until now has been one of the basic assumptions of the programme. The nadir of the approach comes when in the space Amazon story, she concludes 'the system is not the problem'. The real problem, apparently, is violence from the left."

The thing is, if people you're interested in listening to are continuing to use these networks, you've got to use them too to hear those people.