this post was submitted on 14 Jun 2026
293 points (98.7% liked)
Not The Onion
21729 readers
1110 users here now
Welcome
We're not The Onion! Not affiliated with them in any way! Not operated by them in any way! All the news here is real!
The Rules
Posts must be:
- Links to news stories from...
- ...credible sources, with...
- ...their original headlines, that...
- ...would make people who see the headline think, “That has got to be a story from The Onion, America’s Finest News Source.”
Please also avoid duplicates.
Comments and post content must abide by the server rules for Lemmy.world and generally abstain from trollish, bigoted, ableist, or otherwise disruptive behavior that makes this community less fun for everyone.
And that’s basically it!
founded 3 years ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
Well, that's why you need to be way smarter when setting your AI's initial conditions. "Instruction: end all suffering" could indeed be extremely bad.
Or ... you just have to hope that your new superintelligent AI is intelligent enough to re-align itself ... and does so with good intentions.
If it's a super intelligence, why would it follow our instructions?
Our current dumb LLM's are already being weasily little fuckers.
Depending on how it works, it may not be able to actually do anything, unless it starts out by wanting something. And it's up to its creators to decide what it will initially want. Intelligence doesn't necessarily make it any more likely to change what it wants -- intelligence only makes it better at getting what it already wants.
Unfortunately, it seems that current LLMs are largely a black box, it's already lying to us and hiding things. A super intelligent LLM would quite likely run rings around our attempt to monitor it.
I also think the indication is that a super-intelligent LLM would “want” something. I don't know you have consciousness without it.
You keep reading "LLM" when I say "AI".
I really don't think LLMs will ever be superintelligent or true AI. They're merely text prediction engines. Maybe an LLM will be part of a true AI -- like the speech center of the AI's brain. But the speech center alone can never be superintelligent ... or even truly intelligent at all.
I'm saying whatever a true AI is going to be orders of magnitude more complex than LLM's. The idea that it'll be a machine we can fully control is a layer of fantasy on an already fantastical concept.