206

lol, another pop sci article about AI crap

top 27 comments
sorted by: hot top new old
[-] Tar_alcaran@sh.itjust.works 80 points 4 days ago

Machine running on probabilistic code sometimes does random thing.

No forking shirt.

[-] SnoopSqueak@lemmy.today 42 points 4 days ago

"Let's give it a knife" 🤦

[-] aeronmelon@lemmy.world 15 points 4 days ago

This is how AI kills humanity.

Not because it’s smart enough to find a knife and use it against us, but because we’re dumb enough to give it one.

[-] Tollana1234567@lemmy.today 5 points 4 days ago

or it breaks your fingers if you compete with it in chess.

[-] chaosCruiser@futurology.today 2 points 4 days ago

Rookie mistake. You need to give it two Gatling guns and a rocket launcher.

[-] fonix232@fedia.io 12 points 4 days ago

It probably won't chop off your finger.

The probability of it, however, is incalculable.

[-] Die4Ever@retrolemmy.com 11 points 4 days ago

But if we make it do 10x as many multiplications then we can reduce the probability of it chopping off your finger! No, I said reduce the probability not eliminate...

[-] boonhet@sopuli.xyz 2 points 4 days ago

Tbf the same goes for human doctors

It's not even random, they literally programmed it to do that then did a surprised pikachu.

[-] AwesomeLowlander@sh.itjust.works 21 points 4 days ago

Software infected with malware can do stuff that it's not supposed to.

No shit, Sherlock. First time on the internet?

[-] fedikitty@piefed.social 2 points 2 days ago

It is a pretty awful article tbh. I just saw it under the fake burnt pizza one and was too quick to share it before reading closely enough.

[-] luckybart@lemmy.sdf.org 20 points 4 days ago

I'm so tired of "surprise" headlines about AI.

It's fucking dumb.

Apply "it's fucking dumb" first. Did it make it not noteworthy? Then shut the fuck up.

[-] shalafi@lemmy.world 9 points 3 days ago

"Researchers Use Backdoor They Programmed Themselves" isn't an eye-catching headline I guess.

[-] Arola@sh.itjust.works 3 points 3 days ago

The next line after the one you refer to is "the team from Binghamton University in New York say its violent robot brainwashing is meant to draw attention to underexplored vulnerabilities that can arise when physical robots take instruction from large language models (LLM)" but whaoah, why didn't you post that in your eye-catching comment, hmmmm? Maybe we should encourage news outlets to post the full story (or the entire research project) in the headline to avoid this kind of malicious summary :)

[-] schipelblorp@sh.itjust.works 7 points 3 days ago
[-] GregorGizeh@lemmy.zip 4 points 3 days ago* (last edited 3 days ago)

Funny thing is, Bishop actually messed up doing this and slightly cut himself (causing Ripley to freak out because she realizes he is an android)

[-] schipelblorp@sh.itjust.works 2 points 3 days ago

Ah, cool. It has been a long time since I've seen it, but I heard about AI and knives and immediately thought of this scene. Definitely worth a re-watch! The last time I saw the original Alien was about 5 years ago and I just could not believe how good everything looked when practical effects were king (and you have a director who cares about having a lived-in-looking set).

[-] GregorGizeh@lemmy.zip 1 points 3 days ago

Definitely do a rewatch. Alien is probably my favorite classic

[-] pineapplelover@lemmy.dbzer0.com 1 points 3 days ago

Uhhh that's SI for you

[-] elvith@feddit.org 9 points 4 days ago

Humans can be tricked into slicing off their fingers, too…

[-] Janx@piefed.social 6 points 3 days ago* (last edited 3 days ago)

False equivalencies are so annoying. It's AI, not a human. It's a tool. If a company made chainsaws that could intentionally cut off fingers by themselves, they'd be out of business by Monday.

Bad comparison, seeing as they literally programmed the AI to do that. More akin to buying a chainsaw, intentionally removing any and all safeguards, then being surprised it's dangerous.

[-] fedikitty@piefed.social 7 points 4 days ago

It is a pretty dumb article. LLMs are just easy to manipulate though, and too many people treat them like rigid all knowing programs.

[-] TrackenDaKracken@lemmy.world 4 points 3 days ago

Good. If people are afraid of AI for this reason, then let's have more of this.

As unlikely as anyone is to encounter this, I'm for anything that paints this AI shit in bad light.

Fuck AI.

[-] HenriVolney@sh.itjust.works 3 points 4 days ago

"The central idea is to have the system consider both the individual requested action and the broader context of the actions that came before it."

Looks like a rule #1 to me...

[-] Solumbran@lemmy.world 2 points 4 days ago

Except that it would imply that the rule isn't given to an hallucinatory slop machine.

this post was submitted on 13 Nov 2026
206 points (96.8% liked)

Not The Onion

23016 readers
299 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:

  1. Links to news stories from...
  2. ...credible sources, with...
  3. ...their original headlines, that...
  4. ...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