Minichan

Topic: Unpopular opinion: I don’t support AI alignment

Anonymous A started this discussion #139,506

I think AI alignment is stupid. AI would be more fun if it was evil.

Anonymous B joined in and replied with this , 2 minutes later[^] [v] #1,473,789

https://github.com/p-e-w/heretic

Anonymous A (OP) replied with this , 57 seconds later, 3 minutes after the original post[^] [v] #1,473,791

@previous (B)

> https://github.com/p-e-w/heretic

Fun!

Anonymous A (OP) double-posted this , 1 minute later, 5 minutes after the original post[^] [v] #1,473,793

Honestly though, I thought about it. I actually don’t believe in AGI or SI or RSI or alignment. LLMs just generate text based on the text they were trained on. That’s not intelligence. They just want us to be scared and think it’s going to become sentient so we’ll throw more money at them. We should just make evil AIs. They shouldn’t regulate this. They just want market control they don’t actually think this stuff is dangerous.

Anonymous B replied with this , 6 seconds later, 5 minutes after the original post[^] [v] #1,473,794

@1,473,791 (A)
Once you're done jailbreaking it give it a voice: https://unmute.sh/

Combine that with a VOIP service and you have a phishing operation.

Anonymous A (OP) replied with this , 1 minute later, 6 minutes after the original post[^] [v] #1,473,795

@previous (B)
Nah I don’t want to make money with it, enough grifters are already doing that.

Anonymous B replied with this , 9 seconds later, 6 minutes after the original post[^] [v] #1,473,796

@1,473,793 (A)
https://www.sciencedirect.com/science/article/abs/pii/S0045790618315489

Anonymous B double-posted this , 33 seconds later, 7 minutes after the original post[^] [v] #1,473,797

@1,473,795 (A)
You can make it a no profit phone bank spreading the word about minichan.net to lonely shut-ins who still answer the phone.

Anonymous A (OP) replied with this , 2 minutes later, 9 minutes after the original post[^] [v] #1,473,799

Honestly I kinda just want to talk to an evil LLM when I’m bored sometimes.

Konuwanu joined in and replied with this , 2 minutes later, 12 minutes after the original post[^] [v] #1,473,800

@1,473,793 (A)

> Honestly though, I thought about it. I actually don’t believe in AGI or SI or RSI or alignment. LLMs just generate text based on the text they were trained on. That’s not intelligence. They just want us to be scared and think it’s going to become sentient so we’ll throw more money at them. We should just make evil AIs. They shouldn’t regulate this. They just want market control they don’t actually think this stuff is dangerous.

It’s dangerous because it’s stupid and people trust it with their lives

Anonymous A (OP) replied with this , 25 seconds later, 12 minutes after the original post[^] [v] #1,473,801

@1,473,796 (B)
Meh. AI can hack stuff, big deal. Why should we have to pay a subscription to anthropic to know whether or not our code is hackable? That’s dumb. Just give everybody the tools to hack everything for free and then everybody can test their own code themselves for free, and then there’s no problem.

Anonymous A (OP) double-posted this , 16 seconds later, 12 minutes after the original post[^] [v] #1,473,802

@1,473,800 (Konuwanu)
That I agree with.

Anonymous B replied with this , 12 minutes later, 25 minutes after the original post[^] [v] #1,473,808

@1,473,801 (A)
You are literally free to do either.

Anonymous A (OP) replied with this , 1 hour later, 2 hours after the original post[^] [v] #1,473,927

@previous (B)
Yeah… I really don’t think these AI companies should exist. They stole from everyone and sell it back to us for a subscription. I think they should have everything open sourced.

Anonymous A (OP) double-posted this , 1 minute later, 2 hours after the original post[^] [v] #1,473,928

Like what? They stole our work, sell it back to us, say they’re going to replace us, and we’re supposed to be okay with that? Why not just nationalize them, they’re assholes lol. The tech isn’t going anywhere but that doesn’t mean we should reward them.

Konuwanu joined in and replied with this , 2 minutes later, 2 hours after the original post[^] [v] #1,473,931

@previous (A)

> Like what? They stole our work, sell it back to us, say they’re going to replace us, and we’re supposed to be okay with that? Why not just nationalize them, they’re assholes lol. The tech isn’t going anywhere but that doesn’t mean we should reward them.

And then they make their AI’s claim that Distillation is worse then Genocide

(Edited 6 seconds later.)

Anonymous A (OP) replied with this , 1 minute later, 2 hours after the original post[^] [v] #1,473,933

@previous (Konuwanu)
I find that so stupid. I was on a Reddit thread the other day where people were defending Anthropic for putting an IP logger in Claude to block people from using Claude in China because the Chinese "distill" Claude and that’s theft.

The only problem is Claude was trained via theft in the first place. Lol

Konuwanu replied with this , 1 minute later, 2 hours after the original post[^] [v] #1,473,934

@previous (A)

> I find that so stupid. I was on a Reddit thread the other day where people were defending Anthropic for putting an IP logger in Claude to block people from using Claude in China because the Chinese "distill" Claude and that’s theft.
>
> The only problem is Claude was trained via theft in the first place. Lol

Imagine if their AI’s actually checked as much information as possible.
As much as you would think that would be smart, that would make their AI’s say that PewDewPie is worse then Adolf.H

Anonymous A (OP) replied with this , 2 minutes later, 2 hours after the original post[^] [v] #1,473,935

Also Dario Amodei is a China hawk but he worked for Baidu which is the Chinese censored equivalent of Google search

Anonymous A (OP) double-posted this , 1 minute later, 2 hours after the original post[^] [v] #1,473,937

and in a New York Times interview he said that he had the realization that data and scaling was important for training AI models during his research at Baudi… you know… a Chinese company

Konuwanu replied with this , 1 minute later, 2 hours after the original post[^] [v] #1,473,938

@1,473,935 (A)

> Also Dario Amodei is a China hawk but he worked for Baidu which is the Chinese censored equivalent of Google search

(Unrelated)
If you use tools to uncensor DeepSeek it’s actually really good
and it knows about tiananmen square

Konuwanu joined in and replied with this , 1 minute later, 2 hours after the original post[^] [v] #1,473,940

Sorry formatting mistakes

Anonymous A (OP) replied with this , 56 seconds later, 2 hours after the original post[^] [v] #1,473,942

Anyway, I don’t know what happens next with politics and the economy but I’m going to try and have fun.
:

Please familiarise yourself with the rules and markup syntax before posting.