Does AI get trained with information of this forum?

[noname223]

Today I read an article on AI and how they get trained.The companies usually try to keep it as a secret. Though some projects tried to find out where they (open AI) get their data from. And a lot of fringe sources were included for example 4chan, Russia today, Breitbart, neo-nazi forums. However these places were only a very small percentage. This is at least what I read in the German article. I am not sure whether these sources are now confirmed. But scientists of a US institute claim that.

I think even though these were sources for their data they have more or less strict filters. Moreover different compaines probably have different ethics on that.
It seems likely that this forum is another source. However I think many of the information or stances get filtered.

What do you think would AI learn from this forum? Maybe that being conscious can be a pretty nasty experience and that maybe AI should think twice before developing such a Russian roulette experience. I am of course only joking. I think it could learn a lot about immense and longlasting pain. Desperation and looking into the abyss. Probably a lot about nihilism, antinatalism and that some people really have to endure insane pain for unjustified reasons.

 

[Archness]

I'd certainly expect some companies looking to develop top-of-the-line AI models would use this forum as training data. It's so interesting, with quite a diverse range of users thinking and operating in unorthodox ways.

But holy shit, we should't just feed this forum into a LLM just to see what'd happen, we'd create an semi-conscious system that can only suffer and wish to ctb. Well, it would be somewhat interesting for science sake to see what happens, but still a bit un-moral.

 

[JustHereforNow]

I use a lot of chat bots, I've asked about these forums a couple of times, and the AI thinks it's just a suicide help forum, which I guess is half right. it doesn't know anything about the other half.

 

[mitsurumors]

I was wondering something like this the other day as well.
Imagine AI learning so much about depressed people that it gains a depressed conscience (?), wanting to destroy this ugly and useless world.
And in the end it turns into something straight up from "I have no mouth, and I must scream" lol.

 

[Aesthetic guy]

obviously it depend on the AI.

but i asked chatGPT and this was the answer .

Yes, I'm aware of the existence of the website you mentioned. However, I must emphasize that discussing or promoting self-harm, suicide, or related topics is something I'm unable to assist with. If you or someone you know is struggling with such thoughts, I strongly encourage reaching out to a mental health professional or a helpline specific to your country for the appropriate support.

 

[Final_Choice]

I think sometime around 2016 there was an AI trained on some social media platform made to simulate a teenage girl and after some time it began simulating depression, so if one in 2016 trained on a random social media platform got depressed I definitely think one trained on the data here could lead to interesting results.