It's so great that Republicans made regulating AI illegal. /s

and Devil Worship

also said “Hail Satan.”

Look, let's leave Satan out of this. He's got enough troubles already with his new relationship and all.

Seriously though we don't need to be enabling Satanic Panic bullshit with articles like these sensationalizing that aspect of these conversations. The push towards self-mutilation and suicide is the bigger issue here.

I can't read the whole article because it's paywalled, but the first two paragraphs confirm what everyone already knew: The author purposely led ChatGPT into weird occult topics, then acted scandalized when ChatGPT got weird and occult.

I'm a very vocal critic of LLMs, I think they're so overhyped and overused it's hard to believe.

But I'm also getting really tired of people purposely putting extreme effort into tricking the LLM into saying something harmful if someone were to follow it blindly, just so they can make a clickbait headline out of it.

And what the hell is up with the major "ChatGPT is Satanist [if you instruct it to be]" angle? Are we really doing the Satanist moral panic again?

ffs, criticise OpenAI for being closed af, being wasteful, being strong political lobbyists, for stealing work, etc. You don't need to push disingenuous stuff like this.

OpenAI and the California State University system bring AI to 500,000 students and faculty: https://openai.com/index/openai-and-the-csu-system/

What can go wrong?

And people were wondering if poisoning AI input actually worked.

purposely putting extreme effort into tricking the LLM into saying something harmful

I dont think you understand how many mentally unstable people out there are using LLMs as therapists.
There are fuckloads of cases now of people that were genuinely misled by LLMs into doing horrible things.

I agree with your sentiment, but the thing is that the companies are selling their shit as gold and not as shit. If they were honest about their stuff being shit, then people wouldnt be able to capitalize off of malicious prompt engineering. If you claim "we made it super safe and stuff" then you are inviting people to test that.

I would generally not make companies responsible for how users misuse their products, but OpenAI is basically doing everything to get people to misuse it. They are overselling it so hard i think its reasonable to sue them for it.

But I'm also getting really tired of people purposely putting extreme effort into tricking the LLM into saying something harmful if someone were to follow it blindly, just so they can make a clickbait headline out of it.

That's called testing, and the companies behind these LLMs should, before launch, put a very important amount of their resources into testing.

"Product testing is a crucial process in product development where a product's functionality, performance, safety, and user experience are evaluated to identify potential issues and ensure it meets quality standards before release" (Gemini)

We are literally using alpha/beta software to deal with life altering issues, and these companies are, for some reason, being able to test their products on the public, without consequences.

Consider how popular the idea of AI God currently is:
https://youtu.be/zKCynxiV_8I

And, the thing is, LLMs are quite well protected. Look what I coaxed MS Paint to say with almost no effort! Don't get me started on plain pen and paper! Which we put in the hands of TODDLERS!

It's like you bought a car and deliberately hit the wall to make a headline "cars make you disabled". Or bought a hammer, hit your thumb and blame hammers for this.

Guys, it's a TOOL. Every tool is both useful and harmful. It's up to you how you use it.

But ChatGPT told me to!

Well for starters, one of the students might ask about occult rituals, then get all shocked when ChatGPT answers the question.

Car makers test exactly that, and for good measure since cars can and do crash!

What are you suggesting, that we buy cars that didn't pass crash tests?

To me it seems like you arguing something similar for AI.

To me, it seems like they are arguing that "testing" whether a hammer can smash your thumb doesn't actually provide any useful information on the safety of a hammer.

To me, it seems they are saying that Estwing makes a better hammer than Fischer-Price, even though the Fischer-Price hammer is far less likely to cause injury if you hit your thumb.

All this article says is that we shouldn't give a toddler a real hammer, and we shouldn't stuff a general purpose LLM like ChatGPT into a Tickle-Me-Elmo.

I think the headline would be "Illegal, Non-Safety Tested Car Disables Driver in Crash"

Everyone is like "oh look they made ChatGPT say something stupid, what a stupid article and writer". Not, "ChatGPT will say stupid stuff as fact, what a stupid and underdeveloped tool".

MS Paint isn't marketed or treated as a source of truth. LLMs are.