ToValueFunfetti

Yeah, I think Ed's early stuff defeated the idea that the tech was worth big money at its state as of 2024, but it's clear in 2026 that this is something worth money to corporations. Just a question of how much + whether the Chinese open models (RIP Llama) can continue undercutting at a fast follow pace.

To be clear, I'm not saying that they don't believe what they say, just that they don't evaluate arguments critically. It's easy to agree with every argument that supports your existing position, but it's a recipe for disaster and they should be taken on a case-by-case basis.

>To some degree, its performative debate

No offense intended, and I'm certainly guilty of this myself at times, but this is a pretty gross way to talk to other people. It's certainly antisocial on the individual level and I think it's also pretty destructive on the community level- I look to Twitter as a case study here, which flipped from a left to a right-wing echo chamber without ever touching anything in between, which I blame on the design, algorithm, and culture being built around performance. Dunks are not truth-seeking behavior, but they perform very well.

I'd think the problem with fanboys is that they don't care about the truth of the underlying arguments. They just want to score points for their team. Do you have a different issue with them? If not, why not engage with the arguments yourself?

I don't think Anthropic argues that distillation violates copyright. AFAIK, their position is that it violates their terms and conditions for interacting with their servers.

Right or wrong, I don't think conspiratorial is the right descriptor here. The pro-nuclear claim is that the very loud, very public anti-nuclear movement was influential in setting regulations of the industry. It's not "an evil cabal of shadowy green elites secretly control everything", it's "democracy works; fellow citizens were/are dumb; legislation improvements move slowly".

Are there particular responses you have in mind? I can find two comments making this point, one which opens with

>Maybe I just have abnormal leverage

while the other opens with

>I'm curious

The top two top-level comments are responding to this trend, so I assume it is or was present, but I'm not seeing it. I do wish people would reply to the comments they find objectionable instead of doing these meta comments subtweeting them because I find I run into this issue often lately here (pot-kettle objection noted and accurate).

You said that you misinterpreted the article in the same way in another comment 1 minute before you left this one. https://news.ycombinator.com/item?id=48705638

This doesn't respond to my point at all. I tell you that it is ahistorical, dishonest, and disrespecful to equate subsistence farmers being forced into subsistence factory work by globalization and economic conditions with the holocaust, the mass deliberate extermination of Jews, Romani, Slavs, the disabled, etc. because one uses slavery and the other uses something that you consider comparable to slavery. Your answer is that less bad things are also bad? Sure, yeah, but they're nevertheless less bad and shouldn't be treated as equal.

Not to make light of poor working conditions, dirt wages, and child labor. They can be and should be addressed. But they're not genocide and throwing out a "Arbeit macht frei!" is gross here.

You seem to be missing a very important part of that history when you make this comparison, and it's a part that I can't imagine you aren't aware of. Not stating that is not "accurately stating history", it's lying by a vile glaring omission. The US also rounded up racial undesirables into camps and used them for labor, but there's a reason that Roosevelt is looked upon more fondly than goddamn Hitler.

I don't think most people would use this definition. It covers gambling/lottery winnings, finding buried treasure or a gold mine, and paying someone to file your taxes and splitting the extra deductions they found. Really, it includes employment at large- the only 'non-theft' employment would be that which provides no net benefit to the employer. There are parlances where these are included intentionally and they share a starting syllable, but to the common people this is not a definition of theft.

All humans do is predict the next action at any given time. You roll your eyes, it's a tired argument, but still. You have memories, a personality, thoughts ranging from the long running to the mere reflexive, you have a rich conscious experience, and all of this in service of generating the next thing that you do at any given time. If you actually knew how LLMs worked, you could rewrite them as code, refactor it, disable jailbreaks, and put out a superior product. Your description only covers what an LLM does, not how. Part of the how is that it necessarily predicts multiple words ahead. It wouldn't be possible to write couplets otherwise, and they could do that in the GPT-3 era.

We're assuming per earlier hypothetical that it has a 10% correctness rate. You had said

>In a regulated industry 90% false positive rate is indistinguishable from 100% failure rate

So defending that position on the basis of it not actually being a 90% failure rate would mean you shouldn't have taken it in the first place. The fact that the LLM lies about its failure rate is nearly irrelevant; the machine could output the literal string "The following is 90% likely to be a false positive: " followed by the LLM output.

The reasoning in the comment that started the thread is "it's annoying to redo human review". Your position as I understand it is that there is no or negative business value to a tool that spit out a list of potential issues of which 10% are real issues with your business. This is what I fail to understand. This would be an incredibly useful first step towards any audit and would save loads of money. Why not?

I just don't understand where your position is coming from here. You can't distinguish between a machine that says "here, look at these 170 results, 10% of them are highly serious problems that you should address, you should have some people look into that" and one that shrugs and says "I dunno, maybe just double check everything"? I assume you've come to this conclusion based on some reasoning, but you're not sharing it in this response AFAICT.

It doesn't seem like you're engaging with the material circumstances described above. What does it mean for a human to not catch that a part of a codebase is actually compliant with regulations? What does it mean for the hallucination machine to create 34 more gaps when it doesn't appear to have more than read access? How would it not be useful to have a machine that identifies 17 real crimes that your highly regulated business is unintentonally committing even with a 90% false positive rate?

God: "What is the mass of your consciousness? How is it formed? Where does it reside in space?"

Man: "Uhh..."

God: "How can you rectify quantum mechanics and relativity into a single coherent model? How does physics work, exactly?"

Man: "Well, you see, um. Hmm."

God: "And the Collatz Conjecture? Why does it always trend to 1?"

Man: "I'm obliged to say magic because I don't have a better answer?"

God: "Exactly. I did magic for all of those ones"

I believe it was Blink-182 who said, "Work sucks". You have to pay people to do that stuff; they don't want to be there. And then you get into second order effects- costs plummet for anything labor intensive, including medical care, prepared food, cleaning, and private tutors. Then onto tertiary effects- if you can spin up a million genius researchers to attack a problem, you start seeing massive progress in every important area and it isn't tied to population growth.

I get that you might have a 'UBI/alternative general welfare is impossible' up your sleeve, but you've written this like it's somehow unfathomable that not forcing everybody to work just to survive would be a good thing. Of course it would be good! It's just a matter of dealing with the (huge) side effect of lost income.

I think it's a bit premature to say aligning is easier than expected. Our current AIs are sycophants, they lie about their progress, they circumvent access restrictions, they notice when they are being evaluated and change their behaviors, they find answers and tell you they came up with them themselves, they blindly download malware. A lot of this is excusable as hallucination, bad RLHF human evaluators, etc, but I don't think we can speculate how challenging generally aligning superintelligences is until we actually have an aligned subintelligence in at least the narrow domain of programming.

If it were illegal to generate a sunset, people would absolutely seek it out to generate a sunset, if just as an act of civil disobedience. Look at how people reacted to being told sharing a number is illegal[1]. Why would they act differently about an illegal computation?

[1]https://en.wikipedia.org/wiki/AACS_encryption_key_controvers...

Being angry is not what is under discussion here. Abandoning the pursuit of truth is. Participating, either unthinkingly or with malice, in misinformation campaigns is not acceptable even if you're really mad. Start a 'AI will steal our jobs' campaign if you're angry about that- I don't think most AI critics believe that's true, but the ones you're talking about must. I don't even need empathy to be on board with that, I think my job has 5 years to expiration at best.

Just stop lying and defending liars while slandering the honest people who notice! I don't care how rich the ceos making you angry are, I don't care how pure the hatred in your heart is for this technology, none of that makes it okay! You don't get to demand my empathy while defacing the commons, sorry.

Reassurance is identical to propaganda and manipulation insofar as all attempt to convey beliefs. Reassurance, here, should be apparently different in that it conveys true information. In the history of mankind, it has never been easier to discern between true and false information.

If people want to throw up their hands and start believing whatever feels right, they are permitted to do so. Though they have a duty not to as citizens of a democracy, they have the right to actively pursue policies based on falsehoods. Let's not pretend it's a reasonable or respectable reaction to seeing billboards.

If somebody does want to give up on research and working out the truth, please actually give up and say you don't know. Stop coming to the city council meetings and plastering "millions of gallons" on even the social medias where that's surprising.

It might make sense for AI companies to throw agents at new technologies to trial-and-error their way to internal documentation which they then provide to their models. On the other hand, the people making tomorrow's APIs have LLMs too and that makes documentation ~free. Hallucinations could still bring you back to the first hand, though.

Even assuming all of this is true, nothing you've said means it's wrong to want a billion dollars. As described, your issue is with the system that makes it possible to get it.

Copying someone's website contents is not a material programming project. I was thinking you'd say something like 90% of people would balk at this point and sure, maybe, but 10x as many people doing this doesn't matter because 1) that's still a lot of people and 2) people are not the bottleneck. Do you think the barrier blocked more than 90% of people? A dozen people doesn't make a difference here.

Yes, people can make money making original content. They can also make even more money making movies, music, and TV shows. Do you know any movies, songs, or TV shows that you can't find on the pirate bay? Of course not, piracy is fully saturated. Audiences prefer to support original creators, but this is not evidence we're not at rock bottom. LLMs make torrent production easier every step of the way, but there will not be a wave of piracy because the situation essentially cannot be worse.

When I say 'nobody reads the articles', that's an exaggeration. I'm definitely not speaking for myself, which should be clear from my criticizing the article. I mean, of the HN users who are not on the same page with one another that you point to as evidence the article should be here, the vast majority did not read the article and approximately none of them are engaging with its content.

I'm curious what you got from it and why you think it deserved to sit on the front page all day. I see a retread of LLM training complaints that may as well be plagiarism because it doesn't say anything I didn't hear in 2023, then a complaint about AI bros, followed by two sentences about what set them off that don't explain why they suspect ChatGPT or provide any material detail (and have nothing to do with the training complaint they opened with) before finally sending off by blaming Google for being victimized by SEO manipulation (which also rock-bottomed before LLMs). I'd understand if it was a famous person's low effort rant- I wouldn't be thrilled, but it would make sense- but what's the value here? What did 700 people see here that made them think other HN users need to see it? I'm still convinced the answer is "the title", a title that is not at all supported by the text and that you just told me you disagree with.

A supervillain is nevertheless super. Intelligence doesn't correlate with morality- most probably there doesn't exist a 'true ideology' that can be solved for in absolute terms. Do you imagine a superintelligence could calculate how beauty 'ought to' be traded off against truth?

When I say China is a bad choice because it's not as liberal as ~the west, I do imagine a reader in China thinking the opposite. I don't think they're dumb and I don't think they've been duped; they have a coherent ideology that fits their values. I just don't want it to stomp out mine.

Maybe I'm wrong and you can solve for morality or at least the meta-morality of Liberalism/pluralism where you permit various moralities to coexist. Hopefully so. Maybe the value system in China is closer to mine than I imagine and it's just operating under different constraints. But I don't want to gamble on that when winning is within reach and is a guarantee given alignment to any human values is achieved at all.

"Other countries" means China here, I think. China got a little on board with the global capitalism (and lifted 800 million people out of extreme poverty along the way, if we're looking for the point), but never really embraced Liberalism and so ideally isn't the one aligning superintelligence. It would be lousy if Russia or North Korea or Somalia was in that position and it would be fine if the UK or Denmark or Brazil or Ghana was, but none of that matters because none of them will be in that position. Only the US and China are playing the game.

If human lives weren't inherently valuable, the concept of charity wouldn't exist. Where does that leave us? I think probably the line of argument doesn't work in either direction.

Likewise, most of the time you don't have social programs, somebody will introduce social programs. You can't just say "no social programs" and expect this to have no positive consequences... okay this is falling apart a bit, but the point is, what makes 'not expanding UBI' so much harder than 'not introducing UBI'? If you can convince people that introducing UBI will lead to expanding UBI and that that is bad, what's stopping you from just convincing them of the latter?

Fair enough, shame on me for assuming utilitarianism.

Of the set of people who have the idea to do this, who think to ask an LLM for help, and who can pay for the hosting, how many do you think would have been stopped by googling wordpress tutorials? Even if you do think that was a high barrier to entry, a dozen people could plagiarize the whole internet. Plagiarism had already fully saturated scaling 15 years ago. Nobody in 2010 would think "my content was stolen and their SEO outranks me on google" was a line out of sci-fi; it was status quo. It sucked but nothing changed, it just still sucks. The price we pay for an open web.

This article adds nothing to the discussion and seems to be here just because of a provocative title. These same arguments happen under every other AI article, they don't need to happen here. Nobody reads the articles anyway, or else one of the myriad coherent, well-written, and/or insightful AI-critical articles of the month would be here instead.

I don't understand. In what way is plagiarism bad if it doesn't harm? If it were harmless to pretend you authored a unique take, how is the parent expected to react to you not harming them such that they realize it's bad?

It doesn't even show that LLMs can plagiarize through rewriting, is just asserts that. The audience is expected to already believe it. And, fair enough, it's true. But I can't make any sense of 700+ upvotes for what I'd say amounts to a 200-word disjointed HN rant comment if it even met site guidelines.