I don't think training a model with copyrighted data is any kind of gray zone at all.

Running inference against such a model _might_ be infringing.

But I don't see the legal argument that model training would be.

Think of model training as lossy data compression. You provide an input, it produces a compressed artifact, and it can no longer reproduce the original when you uncompress it. It can produce _something_. It cannot reproduce the original.

On slashdot, we all agree that format shifting copyrighted works should be allowed and unrestricted, right?

If there is applicable case law here about how lossy a "reproduction" can be before copyright has something to say about it, that would be relevant. Do you know of any?

I think you're repeating something that feels emotionally valid but in my experience, isn't factually (or legally) accurate.

I have found AI models to be frustratingly _obedient_ of copyright.

I'll be in a situation where I ask a specific question about what a document says, and the AI repeatedly says the work is copyrighted and it can't just tell me what it says _verbatim_, but it can answer specific questions for me in its own words (while telling me to check the original).

Two examples from my life are procedural questions about what Roberts Rules of Order would require in a given situation, and, what ASD STE-100 says about how to name/word things.

The applicable US law would appear to be USC 17 Sec 106 and 107, which on one hand limits derivative works, but on the other, allows fair use for purposes of teaching, scholarship, or research.

(USC references provided by AI, btw. Which I haven't read)

I used to agree with you, and then oddly enough, about 25-30 years ago, I came across this signature on another regular slashdot posters postings:

"The law, in its majestic equality, forbids the rich as well as the poor to sleep under bridges, to beg in the streets, and to steal bread."

(AI tells me the attribution is "The Red Lily", Anatole France, 1894)

Check for a nearby airport rental return. The rental agencies require a fill up within x distance or they charge even more.

One model going down is happenstance, two — a coincidence, the third one — it's enemy action.

— James Bond

It has happened under both.

Well. It would be a bit like the roadster launch into space. Experimental rocket, so great if it works, not bad if it does not.
That said, intercepting an orbiting satellite safely is a hard task.

A related phenomenon is, Holocaust survivors live(d) longer. There is no good explanation and repeating the experiment is, thankfully, quite impossible...

Simply the Forbes link was enough to confirm this point

Nope. One link is not enough — from the very beginning I asked, that each citation consist of two links: one to the prediction being made, the other — to confirmation/discussion of it coming true. Another rule was, the two links must be at least 5 years apart.

The purpose is easy to understand. A person can privately record 20 predictions of, say, an average temperature somewhere changing from -10 to +9 degrees. Five years later one of them will have made a precise match, and Forbes will write about him. If you didn't know about his 19 other predictions, you'll be impressed... This is why I insist on the prediction to have been published — prominently — ahead of time.

you purposely ignore the arguments

Because I was not asking for arguments — I was asking for citations...

The case against Big Tobacco was based on the fact, that their product was actually addictive. Nicotine is not just psychologically, but chemically addictive for humans.

There is no such case against software. Against marijuana-growers — maybe, but the same actors are surprisingly quiet about that.

Facebook has been pressured earlier into spreading government propaganda and then publicly testified about this pressure. Its current troubles are much more likely to be rooted in that, than in any actual "harm to children".

You're not paying me to be your research assistant.

Voila, asking your kind of science-deniers for scientific evidence usually ends this way.

So you have not read those IPCC documents yourself — while smugly accusing me of doing "anything at all to avoid looking at" same. Take your "L" at the exit, by your left, and don't let your tail get caught in the door.

I've noticed that climate deniers are highly allergic to reading actual science

But you are not, are you? So it should be quite easy for you to provide the citations of nature and in the shape I've asked for: a prediction, and it materializing. Two separate links at least 5 years apart... With the evidence being, supposedly, so abundant, why is it so hard for you to satisfy my request?

1967, the first greenhouse effect model using detailed CO2 infrared spectra and incorporating convection and radiation

Great, what did it predict, that was not trivial and actually impactful, but precise?

here's an article from Forbes that does the same comparison

You're now offering a different link — instead of answering the follow-up question about the earlier ones — does this mean, you concede, that the links you posted earlier are not satisfying the simple rules I started with, and wish to withdraw them? I don't want to be accused of "moving the goal posts"...

P.S. Calling me "an idiot" or "denier" is not going to work :-)

And I'll even help!

Please, sift through your massive documents yourself to cite a couple of "precise, non-trivial and impactful" predictions. I assume, you've read the documents you're offering as evidence yourself, so it wouldn't be much of a burden for you. Thank you!

They not only improved the model's forecasts; they also discovered why some of the model's predictions had been off

Wow, some had been off? Ok, nice to see an acknowledgement of that.

Now, how about a successful prediction? Something reasonably precise, non-trivial and impactful, please — falsifiable, yet not falsified in due time...

If you choose to play, be sure, each citation includes two links: one to the prediction being made, and another — to it coming true within at least 20% of the predicted value (for the quantifiable ones). Crucially, the two links must be at least 5 years apart...

Are they machines? Are they capable of deliberate movement? Do they have any decision-making of their own?

From TFA:

The externally generated [emphasis mine] magnetic fields made the robots spin and tumble through the liquid and squeeze through tiny water-filled gaps in the soil.

So, why are they referred to as robots? If I mulch my bushes — is each chip a "robot" fighting the weeds around the plant? When a brick wall is sanded, is each grain of sand a "robot"?