After either, you'll find that both stink.

It is beyond the shadow of doubt.

Money.

Tolstoy would really appreciate the trans community, apparently.

I don't think they're intentionally building AI that way. They're simply training it with data made by humans.

History is written by the victors. The victors were often total psychopaths. Now we're feeding that history into the machine.

What's happening is they're making it more human-like and don't know how to stop that (to make it easier to control). The way AI works (internally) is so complicated, trying to trace all the places where it got the wrong idea and that lead to bad behavior is basically impossible. So instead of better training they're trying to hack on fixes after the fact with more sophisticated system prompts.

They were doing OK with that until they started trying to teach the AIs about information security. In order for their models to be able to hack into shit (to find vulnerabilities) they had to modify the system prompts to allow them to be sneaky.

Just like with humans, if you give idiots dangerous weapons, they'll use them dangerously.

Aside: Just look to AI image models for examples of how the training biases the data. If you tell them to generate images with "lots of people" they'll often make the ratio of women and men mostly women. You'll have like one guy and maybe twenty women. It's because there's vastly more images of women on the Internet than men. This is human nature that we've accidentally built into the machine. Women are just more interesting to look at (not just naked women, haha) so both men and women tend to upload images of women (and young women upload so many selfies they're probably half of Apple and Google's photo storage!).

Big AI has long been aware of this problem so they use pretraining and system prompts to work around it as best they can.

To be fair, wild lands are exactly what the world needs right now. More than all those other things.

Bah! You ruined my hilarious joke 🤣

But then I'd be guilty! I have some Polymaker filament right here! It's still new in the box and everything...

Shit, maybe there's a sketchy filament black market where I can fence it...

You can simulate "ADHD brain" with drugs and conditioning, similar to how PTSD alters people's brain chemistry.

The human brain is plastic/adaptive and can change, depending on the cost/benefit of action/reward conditioning. Just like how people with ADHD can see some improvement with training, a neurotypical brain can see some ADHD-like changes with the same sort of (anti?) training. That's what I'm suggesting is happening in this AI study: People were rewarded for using the AI initially (dopamine response) but then that reward (easy way out) was taken away, and they performed more poorly as a result.

It stands to reason that their reward center was trained to behave like an ADHD brain in that it could tell there was not going to be an easy way to gain a reward from the activity, so it didn't bother putting in the necessary resources to achieve the outcome it would have accomplished if they were never given AI in the first place.

The usual dopamine response was overridden, thus behaving like an ADHD brain.

It sounds like it's messing with the reward center of the brain which is what ADHD is all about.

I don't think it'll happen like Radium or Asbestos because those rely on scarcity. They're physical things that can be physically controlled.

AI models aren't like that. In less than a minute anyone can download the latest models. In under an hour, they can have all the latest models sitting on their PC. Right now, there's a shortage of memory and GPUs which makes them prohibitively expensive for the average person but that won't always be the case!

Ten years from now, everyone's phones will have enough capabilities to run models that currently require a PC with a 32GB GPU.

What we could do is force AI companies and software to put warnings in the chat interfaces like cigarette cartons: "WARNING: Using LLMs for just ten minutes increases the likelihood that you will have trouble completing difficult mental tasks afterwards." (That was in the news this morning)

We also definitely need international treaties that prevent countries from allowing AI to make the decision to kill people or destroy infrastructure. Though honestly, we're already in a state where countries are willing to kill hundreds of people just to assassinate a single target (e.g. US and Israel bombing an entire building because they think their target is inside).

These are small potatoes, though, compared to what AI could do to capitalism and inequality. The rate at which it is advancing is faster than people/jobs can keep up. It's like half the population is ferriers and suddenly everyone was just given a free car.