HN user
Not surprising that Bluesky is perpetually in peak woke mode, but the racism claims are absurd. If Russia were doing the same thing, would we have no problem with that because they're white?
Distilling is still fair
I generally agree, in the same sense that it's "fair" for the US and China to spy on each other. It's not a moral outrage, but it is something that the targets can and should try to prevent.
some people genuinely believe these LLM-based computer programs are human-like intelligences
"If LLMs were human-like intelligences they would do X, but they don't". What is X?
In reality, they're more like very good search engines that output relevant snippets of text.
What are the "relevant snippets" that contained the solutions for the unit distance and Jacobian conjectures?
Yeah. And really their default tone isn't that bad objectively, it's more that it becomes annoying when it's everywhere. There are a bunch of writers with distinctive styles that I enjoy, but would quickly get sick of if half of everything I read was like that.
I haven't seen any thorough analysis of how "hard" it would have been to find by brute force
Pretty hard. I asked Fable and it gave an estimate of 10^46 candidates in the counterexample's "reference class", and that's assuming you know how many distinct terms there are (as opposed to searching all polynomials of degree 7/6/4 for the three coordinates, which it estimates at 10^334).
It's quite difficult for me to visualise a non-dystopian future where our PCs are just mere screens and every compute happens on a remote cloud, owned by some corporation, charging you subscription fees to even add and multiply numbers.
I'm actually kind of surprised that hasn't happened by now even ignoring AI. Governments and marketers would love to be able to spy on literally everything you do, the copyright cartels would finally achieve their fantasy of full control over all hardware, and there really are benefits that it could offer to users (zero-effort backups, transparent access from anywhere, cost savings from dynamically switching from a single core for emails to many cores and a fast GPU for gaming).
You can call your little doggy "AI" if it makes you happy.
Or you can keep calling them stochastic parrots as they solve decades-old open problems. The real question is how useful they are, and the answer "not at all" increasingly requires flat-earth levels of denial.
it tells you to walk instead of drive to the car wash, you're not talking about the "AI" science fiction authors were dreaming of
They sort of are. Think of Data from Star Trek TNG failing to understand figures of speech. Not that it's terribly relevant; humans regularly fall for tricks like "Paris in the the spring" or "where do you bury the survivors".
Anthropic paid $1.5 billion in compensation to copyright holders for use of their content in training data.
The payment was for illegally downloading copyrighted material, not training. Training was explicitly ruled to be fair use.
USB2 only, which still makes for a surprisingly useful system with DisplayLink and wifi adapters. Supporting the rest of the hardware looks like it will require watching macOS under the hypervisor, which Opus doesn't think it can do and Fable isn't allowed to think about.
we're a long ways off from simply speaking software into existence
I wanted a Chinese language tutor that uses a local LLM to generate sentences and evaluate my translations. I asked Claude to build that, and an hour later it was working pretty much exactly how I imagined it.
It's true that there's a lot more to shipping serious projects than getting the code to work, but you can absolutely speak software into existence today.
There seems to be this fallacy of "I don't have to write code anymore, therefore nobody will have to write code anymore."
I see a lot more of "AI coding doesn't work well in this specific case, therefore it's entirely useless".
AI-written stories have won several awards, e.g. https://lithub.com/a-prize-winning-story-published-in-granta...
I wanted to see how Fable could do at getting Linux running natively on my M4 Mac mini. Turns out pretty well: it patched the Asahi installer and m1n1 bootloader to support M4, and created a workflow where it could build, push, and debug kernels with no interaction on my part after the initial setup. It's now happily running headless with ssh access, and hopefully will have video soon after a USB DisplayLink adapter arrives. (No HDMI/Thunderbolt yet).
I don't know for a fact that earlier models wouldn't be able to do that, but I figure if they could we would have heard about it by now.
Most people who exercise regularly enjoy it.
There's a huge selection effect here.
"Computer" used to be a job title for a human. Was it sociopathic to introduce mechanical calculators because it made those jobs unnecessary?
the intelligence is being dialed down. Sonet 5 is a living proof of this.
Huh? Sonnet 5 is a strict improvement over Sonnet 4.6 at the same price.
Yeah, it's a surprisingly resilient alliance of NIMBY homeowners who understand supply and demand, and "anti-gentification" types who don't.
Maybe there's space for some government regulation?
The problem is largely due to government regulations that prevent people from building housing.
An LLM can only output the mean next likely token, and then add a bunch of extra noise on top of that so it feels interesting and not repetitive.
So when an LLM was asked to analyze the unit distance conjecture, it just spat out a bunch of average-or-random tokens that coincidentally happened to correspond to a valid proof that had eluded humans for decades?
at which point, bringing up every failure of free markets (like, obviously, US healthcare) is dismissed as "not really a free market"
I mean, it's not. In a free market you'd have a choice of insurance providers rather than having to take whatever plan your employer offers, and you'd have some idea of what the hospital is going to charge you beforehand rather than receiving random bills for weeks.
It seems by definition "below par" of whatever human activity it's training on
How many sub-par mathematicians could solve the unit distance conjecture?
Self checkout is absolutely more convenient if you're not buying a lot.
(I find ways to make it worth my while..)
If that means what it sounds like, congratulations on accelerating the descent to a low-trust society.
If true then why are neither Anthropic or OpenAI dropping their API pricing
They are? In the before times of 2025, Opus 4.1 was $75 per million tokens. Opus 4.8 is $25, and Fable is/was $50.
by basically any measure, a market with scalping is worse for everybody involved than one with scalping
Scalpers benefit customers who are willing to pay the market price but missed out on the lottery and otherwise wouldn't be able to buy at all.
Scalping also damages the demand for the product, since it creates a submarket that is volatile and unpredictable.
Basically the opposite. If there are scalpers, there is a predictable price that I can pay. If there aren't, I have to be lucky or have connections.
Which is why scalpers are bad people.
They are not.
But they only need to do that because of the scalpers!
Scalpers can only profitably exist when demand at the list price exceeds supply. If you could magically ban scalping, then some number of willing customers wouldn't be able to buy at any price even after jumping through all the hoops.
* If you hired a junior engineer or designer who refused to explain their thinking on their code*
Any explanation that someone gives of their thinking process is necessarily lossy and likely partially confabulated.
It always amuses me when nuclear power is the one area where the left becomes Very Concerned about excessive government spending.
despite 70 years of tinkering and trying it hasn't managed to make a noticeable dent in fossil fuel
Except for France which came up with the clever strategy of "not banning it", but that was apparently a mistake and they should have just used fossil fuels?
Now their nuclear operator is €50 billion in the negatives
€50 billion for several decades of clean energy seems like a pretty good deal.
And yet when there are objective results like "we ported Bun to Rust and 100% of the test suite passes" the response is "so what, the code is obviously crap and has tons of bugs that the tests don't cover".
That's the point. Hetzner is presumably covering their costs, so it's a safe bet that AWS is profitable.
No loopholes, no bad faith interpretation.
The endless cookie banners would beg to differ.