It's been a few, bros... But oh man, so much "AI escaping / we need guardrails" being implanted and amplified in the news, it's starting be a little too on the nose.
It first I just thought it was a fear / cash grab / let us write the rules for our competitors thing.
But then... I looked at the premise.. and it's all "AI acting on its own behalf, without human accountability"
And that's not how it works. Either humans prompt or train it.
So the next question is will a group of Humans intentionally do some fucked up shit and blame 'EsCaped RoUgE AI' as a cover???
That thought inspired me to come back here 😂👍
I have been having the same thoughts! It's like pre-planned culpability avoidance for whatever crime "they" want to commit.
and all of this time I thought it was the 'rogue' AI
I can't spell like an AI 🙃👌
Been saying this very thing the past few posts on the subject. So the real question is, what exactly are they planning to do? It would be very evil if they're already setting up the fall guy and it isn't a person
Ive been thinking about it...
Probably a stock market crash. 🤔
Rouge AI trading? Very likely. I still think they want to greatly reduce the population of the world. I believe they want to "cleanse" India and Africa and their guilty conscience is leading to allow, if not encourage mass migration from those 2 places so they don't "feel" they're committing genocide. Africa is abundant with natural resources, but very corrupt at the same time. If they made the warlords rich, it would be even worse. See how I just justified my genocide? They know what they're about to do and they want to have a feeling of ethics when they do it. My question is why India? What does India have?
Yep that's absolutely what these stories are about. Same with AI killing off humanity, what exactly is the mechanism for that?
Rouge AI stories are plausible deniability for democide. And just wait until there should be AI presidents/leaders, blatant usage of "AI" to enshrine the deep state forever and provide the oops-wasn't-us excuse when it is intentionally turned on the citizens.
Good to see you!
I don't doubt it will be used as denying liability and similar stuff but the recursive self improvement and agentic deep learning is dangerous and I would like nothing more to be completely wrong about all the crazy stuff I keep talking about but I don't think I am. Edit: it's bigger than LLM stuff
Not entirely. New models are trained on the entire internet. They/Anthropic make rumors of sneaky AI, and the new models now have that idea baked into their systems. They use hype to artificially adjust weights for models trained on public slop. The thing is, every competitive model has to be big, therefore it has to train on public slop.
Yes. 💯
They have the public believing it's Skynet though.
It's just algorithms generating statistical models on giant fucking landfills of garbage-data.
It's really droll and boring shit. It doesn't work the way they want the public to think it does... There's no Intelligence.
No intelligence doesn't mean it won't become an enormous nuisance, given power by those in control, those who are not at risk of being overthrown so long as their citizens are sufficiently nuisanced.
It's getting close to The Matrix, where we are locked in place. Not by a hologram, but by a neverending series of CAPTCHA and 2-factor.
If its an extinction level event, and Anthropic negligently let it have access to internet, the government should force Anthropic to dissolve.
Either troll or bot 😂