handoflixue

The "Good" AI is only allowed to draw human attention to an issue - it cannot act, and the humans fundamentally have to (a) figure out what the issue even IS and (b) how to resolve it, basically entirely on their own. Humanity is fundamentally still in charge, the AI just occasionally nudges us to look in the right direction.

Conversely, the "Evil" AI is allowed to do whatever it wants - humanity is no longer in the loop.

And, indeed, the show goes further: they might very well be considered equally moral from some sort of Utilitarian 10,000-foot view. The "Evil" AI wants to save Africa from poverty and pestilence while the "Good" AI is helpless to act on that scale. The moral difference comes down almost entirely to the idea of "minimal intervention" vs "radical revolution"

Then it links to a page which is even more explicit about how little they collect: https://www.infohawk.com/cookies

Certainly the extent of the input matters - "cool art of a dude" is a very generic concept, but at a certain point, just writing the idea down has to count as creative, because otherwise books are all uncreative.

A Photograph is, mechanically, 100% machine-created. But we still call photographers "creative" for their work on picking what to photograph, how to frame it, timing the shot, etc..

Honestly, I think it's wonderful that editors are finally getting some credit for the creativity they bring to the process. There's plenty of writing about how famous films like "Star Wars" often owe a lot to their editing, even though that traditionally doesn't get credited as "creative" work.

My understanding is that the GPU market is largely sold out for the next few years?

What exactly is the threat here? Like, which specific competitors, what efficiency improvements, what alternate platforms?

Most of the October CVEs we've seen were situations that require an extremely complex chain of attacks - amazing if you want to break into a bank, or fuck up some grocery freezers, not so useful for scanning random public boxes and doing "hacks of opportunity"

Apologies for the lack of links - it's a scattering of random people I've read talking about this, not any coordinated study. But if you read between the lines on all of the threads about these CVEs, you'll notice a lack of "I was hit by this" stories - which is pretty normal for more complex vulnerabilities.

The reason these sat unnoticed for so long is precisely because it's difficult to actually exploit them.

If anyone has anecdotes to the contrary, I'm definitely interested in hearing it, of course!

Maybe I'm unfamiliar with the local law, but it sounds like you're just making stuff up, since you're not citing any specific code here.

I suppose if the robot literally cannot path around someone standing still, it might be a valid excuse? But as soon as you're actively moving to block it, it is obviously obstruction.

Judges get to run off common sense, not programmer logic - the point of the law is that public paths ought to be usable by anyone for lawful purposes, and currently "delivery robot" is a lawful purpose.

The whole point of a tool is that it increases the amount of output we can expect from workers. Otherwise we'd still all be doing agriculture, because the farmer's bill of rights sets the maximum output of a farm and there's not enough surplus for non-farmers.

A lot of the recent major exploits like php2shell require a fair amount of effort to use, so unless you're a major public figure or holding on to something very valuable, there's not a lot of reason to be concerned.

Obviously, if you're writing a password manager, it's a very different story and I'd be horrified by such a policy there. But you've got to make an argument that this specific app is actually a target

You said it yourself - the enforceable policy is a a different, and much more reasonable policy. If the rule is "ultimately" something else, you can in fact just make that the rule. Or include a parenthetical explanation.

Rules are the ideal place for being pedantic and detailed! That's the whole point of rules - ultimately, the rule is the rule that was actually written down. Not anything else.

It's not hypothetical to say that an LLM can write widely adopted, battle-tested, production code.

Claude Code would be another example.

Linux is still mostly human, but even Linus Torvalds has warmed up to the idea that LLMs are tools a skilled user can use to contribute there.

There's a lot of quieter, private examples where some company rewrote their entire back-end or internal tooling with an LLM. You just tend not to hear about it because "we used AI" is currently negative marketing/PR, so why say it out loud?

Yeah, but that doesn't imply equivalent exchange.

If you care about being the best sprinter in the world, yeah, put in the time and effort! But if you just want to get to work, there's no way you can beat the speed of a car.

If you're one of the better bug-hunters in the world, you probably lose something by using an LLM. For the rest of us, it's increasingly like trying to race a car.

We definitely need both types - it's important having the humans who do have that top 1% understanding. But increasingly, for everyone else, LLMs outperform doing it by hand.

I really wish a lot more of this debate could show that level of respect for the other side.

Humans also get details wrong when they do stuff like this. So right away, we learn this isn't a limitation of AI. If you want to say that getting that wrong is "proof they don't have a world model", that cuts both ways - it means humans do just fine despite not having one of those either.

But that also lends us insight into fixing the issue on merit! If humans also obviously lack a world model, how do we cope with this? Sometimes we use "live action" and reality forces consistency. Sometimes we use editors and QA. Sometimes it ends up on a website listing every little shot-to-shot inconsistency in every single TV show and movie ever made.

And now you have a few clear ideas on how to train the next generation of image models to do better! And you can probably dramatically improve your own prompting by providing reference images, using consistent descriptions, etc..

So, hardly irrelevant.

It's a pretty short article compared to what usually gets posted here.

Like, technically these requests cost nothing, because they pass the costs on - and because in this case, they've presented an absurd cost to avoid having to do anything at all

You can only make a FOIA request to the US government.

I should think it obvious that yes, every department should in fact be staffed to a level that allows them to comply with this (and all other) legal obligations. Doing otherwise is obviously malicious.

This law was passed last century, back in 1966. It's not like this is some surprising new obligation that they failed to staff against.

"Must declare any purchase of 10 or more GPUs" does not strike me as particularly implausible.

There's already plenty of jurisdictions where possession of certain tools is considered evidence of criminal intent.

I don't think any of this is good behavior on the part of our legal system, but I do not doubt the legal system's ability to regulate this if it so chooses - and Anthropic has already gotten slapped down by the Trump administration once before.

Very different from a single-player video game, or a website with a login form.

At a dead minimum, I'd expect something like "addressing the concerns the experts are raising" instead of trying to do a character assassination.

I generally read "character assassinations" as suggesting that someone has no real technical arguments available - conversely, the experts here are citing specific facts and evidence.

For example, it would be tiresome to have someone insist that s0ss is clearly an illegal hacker with multiple felonies to his name solely because he chats with YCombinator's secret anonymous hacking collective. And yet this post right here makes it very clear you are one of us! ;-)