That's how you get a culture against cheating. You ensure that cheating doesn't pay, and eventually people learn that cheating doesn't pay. The enforcement is part of the culture.

Not saying AI is blameless, but I'm seeing a trend where the problem is clearly more about social media and how it enables every one of us to live in our own algorithmically fed bubble. Like, look at this:

> If your initial reaction to reading that and seeing [an AI image] is some variation of "ughhh" or rolling your eyes or "fuck this guy" congrats. You are normal.

> If it wasn't I cannot stress to you enough that you are an outlier. Whenever you pick key art for a presentation or blog, your business, or whatever - if you use AI art you give a clear signal that you have low social literacy. You immediately associate yourself with a huge bundle of negative emotions because people, largely, hate this shit.

See how confident the author is that their own view is the normal, socially acceptable one, and they "cannot stress enough" that any other view is a social outlier. This has all the same "If you don't care about what's happening in Gaza you're not normal and nobody likes you" energy, except at least people are actually dying in Gaza.

And of course it should make perfect sense for the author because, unironically, everywhere they go online they will see people talking and thinking like that! Largely thanks to those profit-driven corporations and their massive data centers.

I don't know what's the solution but this can't go on forever ...

Some data was permanently lost, and then officers told reporters that multi-regional backup was not yet built because it was too hard at such a massive scale... of 858 TB.

I mean, what's Trump going to do? Murder Iranian leaders harder?

I have a feeling that if LLMs were built on a deterministic technology, a lot of the current AI-is-not-intelligent crowd would be saying "These LLMs can only generate one answer given a question, which means they lack human creativity and they'll never be intelligent!"

Well, to me, this sounds basically like "Jeff Bezos already exists, this school does not. Increasing the government budget to build a school here is not going to help our finance, so that's where we will push back."

(I don't think Jeff Bezos should lose all his money, but he could definitely pay more tax.)

And it's good that we can think that way, because we also follow the rules of statistical and quantum physics, which are inherently probabilistic. So, basically, you can say the same things about people. There's a nonzero (but extremely small) probability that I'll suddenly go mad and stab the next person. There's a nonzero (but even smaller) probability that I'll spontaneously erupt into a cloud of lethal pathogen that will destroy humanity. Yada yada.

Yet, nobody builds houses under the assumption that one of the occupants would transform into a lethal cloud, and for good reason.

Yes, it does sound a bit more absurd when we apply it to humans. But the underlying principle is very similar.

(I think this will be my last comment here because I'm just repeating myself.)

LLM generating each token probabilistically does not mean there's a realistic chance of generating any random stuff, where we can define "realistic" as "If we transform the whole known universe into data centers and run this model until the heat death of the universe, we will encounter it at least once."

Of course that does not mean LLMs are infallible. It fails all the time! But you can't explain it as a fundamental shortcoming of a probabilistic structure: that's not a logical argument.

Or, back to the original discussion, the fact that this one particular LLM generated a command to delete the database is not a fundamental shortcoming of LLM architecture. It's just a shortcoming of LLMs we currently have.

And I do think it's stupid to wire an LLM to a production database. Modern LLMs aren't that reliable (at least not yet), and the cost-benefit tradeoff does not make sense. (What do you even gain by doing that?)

However, you can't just look at that and say "Duh, this setup is bound to fail, because LLMs can generate every arbitrary sequence of tokens." That's a wrong explanation, and shows a misunderstanding of how LLMs (and probability) work.

This is just trivially wrong that I don't understand why people repeat it. There are many valid criticisms of LLM (especially the LLMs we currently have), this isn't one of them.

It's akin to saying that every molecules behave randomly according to statistical physics, so you should expect your ceiling to spontaneously disintegrate any day, and if you find yourself under the rubble one day it's just a consequence of basic physics.

After all, MAGA supports Trump no matter what. ICE murdered two US citizens already. Trump could shoot a baby in Times Square and I don't think it will ever go below 30%.

But once you raise gas prices, that affects everyone's living cost. I don't think whatever video Epstein made could ever match that. (And these days MAGA will wave it away with "guh it's clearly AI, you libtards." You can't do that at gas station.)

...It's more likely than you think !!

"Let's give the money back to voters because they will like that and we'll figure out something else in tough years" is, like, the quintessential example of "raiding these coffers."

It's basically like big tech companies turning profit into stock dividends because investors love it and the CEO will be handsomely rewarded, and who cares about long-term R&D. When big companies do that we blame MBAs and capitalism.

If the US government neglects a section of highway until a city becomes unreachable by roads, there will be riots. The same city losing a train service? Totally expected, trains are supposed to suck.

The sorry state of American public transport is a self-fulfilling prophecy: everybody knows that public transportation sucks, and therefore nothing is done to improve it, because it's a waste of resource.