News for hackers. Stuff that matters.
Re:Claude Fable 5.2 solves two open problems in the physics ... by dap on Thursday November 19, 2026 @03:05PM (#50565562)
"Claude solves" feels reflective of the constant, problematic anthropomorphizing of the machine. For how it's problematic, look at "it broke out of the sandbox" vs. "we failed to build an adequate sandbox".

What does it mean for Claude to solve it? What was the prompt? I didn't read it all, but the word "prompt" does not appear in the blog post, either abstract, or either paper.

Re:Claude Fable 5.2 solves two open problems in the physics ... by tomrod on Thursday November 19, 2026 @03:07PM (#50565599)
It means someone set up the question with the Claude tool and let it iterate until it came up with an acceptable solution.

When my accountant whips out a calculator, it is the accountant that defined the problem and set the solution in motion. Fantastic tools are still tools. Claude does not love me nor you no matter how its trained and encoded patterns express.

Re:Claude Fable 5.2 solves two open problems in the physics ... by KoolKat23 on Thursday November 19, 2026 @03:19PM (#50565766)
When you ask it to make a banana picture and it goes of on a tangent solving a black hole physics problem, what then.
Re:Claude Fable 5.2 solves two open problems in the physics ... by tomrod on Thursday November 19, 2026 @03:46PM (#50566117)
Congrats on the tangent!
Re:Claude Fable 5.2 solves two open problems in the physics ... by garciasn on Thursday November 19, 2026 @03:47PM (#50566141)
Michael would be pissed it wasted $10 in tokens?
Re:Claude Fable 5.2 solves two open problems in the physics ... by geraneum on Thursday November 19, 2026 @06:04PM (#50567820)
> It means someone set up the question with the Claude tool and let it iterate until it came up with an acceptable solution.

And you know the details how?

Re:Claude Fable 5.2 solves two open problems in the physics ... by utopiah on Friday November 20, 2026 @01:05AM (#50570760)
That's definitely not how it happened anyway. It wasn't until "it" came up with an acceptable solution but rather when whoever prompted it do so felt it was good enough according to whatever criteria they thought was sufficient. Knowing when to stop is in itself extremely challenging so unless there is an objective way to verify, which I doubt it's the case when it's the state of the art, surely someone went through the output and stopped the process, namely no more iteration (which again does show involvement from qualified someone).
Re:Claude Fable 5.2 solves two open problems in the physics ... by dap on Thursday November 19, 2026 @07:48PM (#50568738)
Right, but in that situation, the email from the accountant isn't titled "TI-34 finished your taxes". The accountant did all the interesting work. (This metaphor is getting stretched, since accountants usually use software to do the taxes, and that software does a lot more of the interesting work than a pocket calculator does. But even then, the accountant has done the most important work in telling the software about your tax situation and the summary is still "the accountant has finished your taxes".)
Re:Claude Fable 5.2 solves two open problems in the physics ... by water42 on Friday November 20, 2026 @03:04AM (#50571567)
Calculators don't run on their own with a vague input.
Re:Claude Fable 5.2 solves two open problems in the physics ... by 2OEH8eoCRo0 on Thursday November 19, 2026 @03:21PM (#50565789)
It probably means some physicists were using it to study black holes and it stole their work.
Re:Claude Fable 5.2 solves two open problems in the physics ... by j-pb on Thursday November 19, 2026 @03:23PM (#50565825)
Oh shut up, nobody cares about this constant whining about "the models are bing anthropomorphized too much". They work better when you pretend they are just another human, of it's weird and it works it's not weird.
Re:Claude Fable 5.2 solves two open problems in the physics ... by make3 on Thursday November 19, 2026 @04:37PM (#50566787)
it's also meaningful though, in that it's a tool trained on enormous amounts of stolen human thought. so recognizing that it's the stolen thing that did the thing, and not the company itself, gives people back some of the credit
Re:Claude Fable 5.2 solves two open problems in the physics ... by 00dazzle on Thursday November 19, 2026 @05:25PM (#50567348)
"Spinning Jenny-spun yarn" feels reflective of the constant, problematic anthropomorphizing of the machine. For how it's problematic, look at "the boiled exploded" vs. we failed to build an adequate boiler".

What does it mean for Jenny to spin it? Who provided the power to it? I didn't read it all, but the word "power" does not appear in the patent, or the newspaper articles about it.

Re:Claude Fable 5.2 solves two open problems in the physics ... by dap on Thursday November 19, 2026 @08:17PM (#50568943)
I'm confused by this. Are you comparing an LLM to the spinning wheel? In that case wouldn't the analogous headline be "Spinning wheel spins skein" or something? And if there were an article about it, I would definitely expect it to explain that Jenny operated the wheel and some basics about the steps involved. I don't know enough about spinning to know if it's purely mechanical or if operator skill is involved. If spinning yarn is mechanical, then it's not analogous because it's not mechanical to get Claude to prove a novel result in physics. If spinning yarn with a wheel involves a lot of operator skill, then I wouldn't expect us to say "the wheel spun the yarn", but "Jenny spun the yarn [aided by her spinning wheel]".

To pick yet another example: you _can_ say "the car drove from California to Florida", but that'd an unusual stylistic choice and so would probably imply that you didn't know for sure who drove it or something. In ordinary conversation people would be much more likely to say "Jenny drove from California to Florida [in her car]".

Is the boiled/boiler thing related or just some other example? That isn't analogous either. The analog would be "the pressurized gases escaped the vessel" vs. "the vessel failed to contain the pressurized gases". These are different. If you want to emphasize how out-of-spec the internal conditions were, you say "the pressured gases escaped the vessel". If you want to emphasize that the chosen vessel was obviously wrong for the job, you say "the vessel failed to contain the pressurized gases". Either way though there's no risk of the reader anthropomorphizing the pressurized gases. That's different from "the agents escaped the sandbox" because that implies ... agency.

Sorry if I've misunderstood!

Re:Claude Fable 5.2 solves two open problems in the physics ... by dap on Friday November 20, 2026 @10:50AM (#50576532)
lol, oops, spinning Jenny
Re:Claude Fable 5.2 solves two open problems in the physics ... by jibalt on Thursday November 19, 2026 @08:24PM (#50568985)
It's not at all problematic for English speakers, and this complaint obscures actual problematic anthropomorphizing.

> What was the prompt?

Find open physics problems and solve them, basically.

Re:Claude Fable 5.2 solves two open problems in the physics ... by snayan on Friday November 20, 2026 @01:07AM (#50570774)
I feel like this is a bit of a stretch. The problem was solved by who or what if not Claude?
Re:Claude Fable 5.2 solves two open problems in the physics ... by shrx on Friday November 20, 2026 @02:39AM (#50571383)
By the person interacting with claude.

By the persons who trained the model.

By the persons who provided the data the model was trained on.

etc.

Re:Claude Fable 5.2 solves two open problems in the physics ... by snayan on Friday November 20, 2026 @04:24AM (#50572209)
Agree to disagree, I don't think it makes sense to say any of the options you listed solved it.