Ironically, SELinux might be a pretty good tool to keep LLMs from accidentally getting the answers to their benchmark

OK 👍

Mention you just used AI with the wrong people and you’re out

So, there are people who think you are a bad person for supporting large companies with fascist owners that sell you a product that makes you unlearn critical thinking, steals your data and also destroys the environment in new and funny ways?

Yeah, that sounds like a reasonable opinion to me, to be honest. Of course, you are free to disagree, but I would like to use what is left of my free speech to call AI supporters morons. I am also deeply against people who make cluster munitions to be dropped on middle-eastern hospitals, or who work in surveillance capitalism.

That’s just me having an opinion, and I will not give up my core values because of some hypothetical unity. That is how we end up with the US Democrats being the “left” party. People like Clinton or Pelosi have tend to use arguments to silence critics, and it didn’t work out well for the US left. The enemy of my enemy is my ally, not my friend.

I am not a lawyer, this is not legal advice, as should be obvious by me being a random dude on the internet.

• you can kind of sue about breaches of contract, but that’s civil law, not criminal law, so not illegal
• unless I profit from that or use it to defame <redacted>, they might have a hard time demonstrating their damages from that breach
• I didn’t sign any contract with <redacted> over their firmware anyway, and it will be hard to prove for them that I did, so kiss my ass
• I’m not in the US, neither is <redacted> so the DMCA can kiss my ass as well
• there are people who do this kind of work above-board, mostly in academic contexts, with their name attached to it, and as far as I can see they still walk around freely

That line of thinking is directly leading to “you cannot say anything any more” and “facts before feelings”, and before you know it, you confuse some loud people disliking your edgy opinions with you and your “facts” being oppressed. It’s human interaction, get over it, stop being butthurt and either engage in constructive discussion, willing to occasionally accept being wrong, or block and move on. Most of us live in a world where we have more freedom of speech than most of recorded history. Stop perpetuating that bullshit play by people who hate people haveing civil discussions, because they profit from screaming matches and endless polarization. Someone disagreeing with you, even someone calling you an idiot, is just exercising their free speech, same as you do.

With that out of the way, several things can be true at once: intelligence and reasoning can be ill-defined terms, an LLM can be able to match inputs to outputs with impressive accuracy, it can have an internal model of the world that is very weird and surprisingly lacking, and it can be a dangerous technology for humanity. Does that make it intelligent? Probably, for some definition of intelligence. But I am still much more worried about people with/behind AI than about AI itself. I’m not afraid of a gun, I am afraid of an egoisitic moron with a gun.

That “moron” was not directed at you, but only at my coworker. I just wanted to point out that it seems to be a bit hit-or-miss sometimes :)

Apparently you write good meeting notes.

Don’t get me wrong, the models are really impressive, and something I would have called impossible two years ago. From an information theory point of view, it is super cool how they manage to compress so much knowledge and connections in so little size (relatively speaking).

Itelligence is very ill defined, as someone else pointed out. It still has a very weird model of the world (if at all).

My point is however, it sometimes seems to be too good at guessing facts that should not be guessable from the inputs and publically available information alone.

A coworker did something similar recently. Besides pasting internal stuff into a foreign model like a moron, it had some nice ideas, but lacked a lot of information it simply could not have known. Things like, we already did this, but in an unrelated repository, and we didn’t mention it in this part of the docs. Or, since the person did not know / input the overall purpose and constrains correctly, the LLM proposal was cool, but missed the point completely. Or, the LLM forgot to mention that its plan would require buying hardware we simply don’t have yet, and which would be really expensive and hard to obtain right now, because, idk, hallucinations or it just assumed it was all already there.

Yeah, not impressed. Neither by that idiotic coworker nor by the proposal.

Nah, I don’t think so. We’re getting into philosophical / religious territory here, but from a technical standpoint I believe there is no difference between a brain and an accurate simulation of a brain, and an accurate simulation of a brain is “just” an engineering challenge at this point. We might now have the computing power or architecture yet, but there is no conceptual issue prohibiting us from building it.

But an LLM is not a simulation of a brain. It is an autocomplete on steroids. Very different architecture. It is not a brain, but it can sometimes produce the same outputs a brain would. But it is nowhere close to a human brain.

Which makes it a breach of contract, but not illegal.

We steal your data: $5 We steal your data and tell you we don’t: $10

Absolutely. Especially if you have company info / IP that someone else would pay a lot for. This is a question for lawyers and managers to negotiate with the lawyers at OpenAI, Anthropic etc, not for us little people to decide. I’m not throwing company data in an online service that is not cleared for us to use, and I don’t even work with spicy stuff. Back in my IT days we would have called that a data breach, which would have lead to something from an unpleasant hour in the boss’ office to a promotion to customer.

Better at replicating closed source software.

It surely gets better at coding in general, after all there are smart people at anthropic with near limitless budgets, but my conspiracy thesis is that it gets a lot better at replicating stuff it has seen before, but really shouldn’t have.

Tinfoil hat time:

A year ago, I did some reverse engineering. I figured a pattern matching engine has to be good matching patterns, so I threw the messed-up output of Ghidra into Claude to clean it up and to give me an overview of what the fuck happens in a certain part.

I was absolutely surprised. Sure, it got some things wrong, but it knew way more than it should have been able to guess from that short bit of fucked up, optimised not-quite-C. Including guessing context that it should not be able to guess. This was a firmware with not-yet-released functions that, to my knowledge, shouldn’t exist outside of this bit of binary firmware. At least not for us non-NDA mortals.

At the time, Mistral had a new-ish model as well, so I figured I’d give it a try. It worked as expected, cleaned up things, gave variables reasonable names, made educated guesses and so on, but didn’t act as if it had background info. And I was using Sonnet 4.x, so the small Claude which shouldn’t be that much larger.

Since then, I have the conspiracy theory that the Claude people threw a bunch of proprietary sources in the training, too. I have no idea how, maybe aggressively training on user inputs? Anyway, this does not help to allalleviate my suspicions.

Yeah, it is not an age thing. I’m very happy that my mother has always been pretty good at programming solving. Although she never did any real tech related job, I believe this is a question of attitude more than anything else.

When she proudly tells me she used my dad’s phone to google for a fix for her phone acting up, my inner first year support monkey is proud :)

I remember a large company picking MS over Google Docs, because at least at the time, MS proudly advertised that you could show up with a wagon full of HDDs at their data centre and get your stored stuff. Although, a lot of enshittification happened since then.

Nah, that would be productive and not flashy enough for a publicity stunt.

Same. My 60+ mother got a new computer recently, saw the prices of MS Office and told me “install that free thing”. So far no complaints, first remark was “well, looks like it has all the same buttons, works for me”

The way I unterstand it, at least in European practice, the code itself is copyrighted, as in, the actual text, not the concept behing it. Decompiled source is not the same as original source, it goes through rounds of compiler optimisations and the process of reversing that includes a lot of guesswork. So, in practice, decompiling and working on the decompiled source was at least tolerated, probably because it would have been a grey area at worst.

Hey! We need to save water for starving LLMs!