David Shipley

It's supposed to be a sycophant by design.

And they were able to use these well-known techniques from door-to-door salesmen from decades ago to convince these AI to do really bad things, including generate recipes for certain drugs.

Exactly, but imagine...

if the recipes, the code recipes that this thing was taught included really good recipes as well as really, really terrible recipes.

And it does not know the difference.

In fact, the more often something appears, like a common coding error, the more likely it is to think, this is what you do in this sequence.

So it literally took bad code

and made it the median, the average code that this thing produced.

So the OWASP top 10, if we're really nerding out on the show, there's like 10 known coding screw ups that everyone still keeps doing today.

And this thing literally was trained on code examples that have a lot of this in it.

So it often produces this.

It's meant to be a power saw, not a self-driving car.

Well, I mean, there's a lot of bad things hidden in LMs, not just bad code.

By the way, the amount of racist, horrible, misogynist, anti-human, misanthropic content hidden inside these models.

Imagine a mental model for sake of these AIs.

They're like a landfill for a large city.

And eventually you cap it, you put a park on it, and you're trying to just get the best of the park, but you catch a really bad smell every now and then.

That's all the awfulness that's hidden that they never dealt with.

That is the TLDR.

This was a super bad idea, and yeah.