Lack of evidence is not evidence for your pet theory.

Seems like it’s hard to find the actual repository and they put it behind a trigger warninf: https://github.com/terrafying/ai-torture-chamber

Here’s an idea: don’t use unsafe software where safety is needed.

I’m sure the AI told them they were right to push back when they confronted it

I mean it doesn’t help she went republican lite on immigration.

Cthulu is an archetype, right? Comes into our world with no regard other than what it can manipulate and drives people insane.

Edit: thinking about it, it’s not an archetype when it comes from an LLM. The consistency it does display doesn’t come from some conviction an entity should be a certain way. Ask an LLM to put an archetype in a novel situation and it falls back to matching forms.

I appreciate the emphasis on conviviality or human centric design. But boy I want to nitpick on the following:

when we collaborate with AI systems, we might be interfacing with the archetypal patterns that have guided human wisdom throughout history. This makes programming with AI a form of accessing collective spiritual wisdom rather than just advanced tooling.

I agree with the first half, that yes, you are engaging with archetypal patterns that is often present in human wisdom but that is where it stops. There is no guarantee the LLM has selected the appropriate pattern for your use case. There is no wisdom from the LLM because there is no conviction from its choices. The LLM, at best, is a superstitious machine that sees that X pattern worked in a situation with similar signifiers. Wisdom is understanding which signifiers are of consequence given your context, which requires being able to ignore/forget irrelevant context - something an LLM is terrible at doing as evident by all the “context engineering” tools that exist. It feels like a chicken and egg problem where in order for the LLM to “get right” what is the appropriate context it must first have the appropriate context.

“I was aiming at the bird that was attacking me”

Heard a rumor that one of his advisor is married to an AI specialist at Google. Like I don’t expect Sanders to know who yud is unless someone on his staff told him it would be a good idea to talk to him.

Who is advising Sanders on AI?

** Treaties made with the Natives are subject to black out dates

But I can’t afford to buy a congress person.

When has 196 ever posted anything not satire?

The only notes are the ones I’m taking.

Now what would a group of horny middle schoolers do to a bot like that?

Is that suppose to impress us? I too can copy code

My first set was aluminum, and I got it from IKEA

Good point. Now I imagine someone trying to pass the turing test by having their bot pretend to be a sales person.

I guess a certain country is using LLMs to try to engage people 1:1 to change their opinions of said country.

Came across this today: the purpose of a system is what it does. Based on that I would say the purpose of Claude was to make mistakes and get my coworker fired.