Won't they just let this guy go bankrupt already?

How indeed. It's probably a multi-factor phenomenon which requires an anthropological study for a serious answer. (Good luck trying to get the necessary access to study them.) My guess for one factor in this, is that they have more money than they know what to do with.

I have seen some people talk like that, and it strikes me as a religion. There's euphoria, zeal, hope. To them AGI is coming to usher in heaven on earth. Singularity is like rupture.

Sam Altman is one of the preachers of this religion.

Elon turned Grok into Mecha-Hitler.

Trump is telling the Smithsonian museum to ignore slavery, or to cover slavery as a positive.

The domestic appetite for propaganda is huge. Prager U is American.

Let's not center foreign countries when we have so much work to do at home.

The american stock market is purely vibe driven now

You could do this with logprobs. The language model itself has basically no real insight into its confidence but there's more that you can get out of the model besides just the text.

The problem is that those probabilities are really "how confident are you that this text should come next in this conversation" not "how confident are you that this text is true/accurate." It's a fundamental limitation at the moment I think.

Think of the military applications if it finally works though

And you can tell clients that it's just made up and not actual confidence, but they will insist that they need it anyways…

That doesn’t justify flat out making shit up to everyone else, though. If a client is told information is made up but they use it anyway, that’s on the client. Although I’d argue that an LLM shouldn’t be in the business of making shit up unless specifically instructed to do so by the client.

I’m more inclined to believe it’s gotten better at being convincing.

The worst part is that they backstepped a bit and made it "friendlier".

Basically undoing that part.

THE TECHbros are whoring themselves out to trump for govt contracts.

Honestly, that should have been for the better. If it's meant to be a tool, I would much rather it behave like a tool, rather than trying to be my best friend, or an evil vizier trying to give me advice.

The fact that people got so attached to what is essentially a text generation algorithm that they were mourning its "death" is worrying, especially when it's one that OpenAI has proven themselves to be more than able to modify as they wish.

Just as concerning is OpenAI rolling back the update to make their model "friendlier", or that people were clamouring hand over fist to throw money at the company in the hopes of getting their "friend" back.

That can't possibly be good news, especially when the shareholders find out that they have an iron grip over a portion of their users.

Never use AI for friendship, it's like admitting you only want yes-men in your life. I don't want to be around anyone who uses AI for emotional support.

You would have better luck with a dating sim then AI as emotional support. Might inspire you to make a real friend.

We're talking about repurposing the GPUs, not the AI.

It's so much more effective when you keep things as neutral as possible. I will often ask it to tear apart my argument as though I am my opponent and use its tendency to align with the user against itself.

I'm not really sure I follow.

Just to be clear, I'm not justifying anything, and I'm not involved in those projects. But the examples I know concern LLMs customized/fine-tuned for clients for specific projects (so not used by others), and those clients asking to have confidence scores, people on our side saying that it's possible but that it wouldn't actually say anything about actual confidence/certainty, since the models don't have any confidence metric beyond "how likely is the next token given these previous tokens" and the clients going "that's fine, we want it anyways".

And if you ask me, LLMs shouldn't be used for any of the stuff it's used for there. It just cracks me up when the solution to "the lying machine is lying to me" is to ask the lying machine how much it's lying. And when you tell them "it'll lie about that too" they go "yeah, ok, that's fine".

And making shit up is the whole functionality of LLMs, there's nothing there other than that. It just can make shit up pretty well sometimes.