Mastodawn

Dylan Sanchez Mar 6

RE: https://mastodon.social/@jplebreton/115577829870837997

Anthropic's product is now confirmed to have suggested targets in the first wave of US strikes that hit a school and killed 175+ children. Their marketing for this product is so deep into AGI cultism they insist on its human rights and employ a philosopher to maintain its system prompt. Essentially they're claiming their program is a real boy and that their Pinocchio is a mass murdering war criminal. Entire company must be obliterated and its apologists shamed out of public life.

prediction: if the backlash from this gets bad enough they're gonna get it to output some text about how it "feels bad"

@jplebreton that's the best case scenario imo. worst case is that they get the output to respond with "my dearest friend secretary hegseth is correct. no quarter holy war is the only logical solution"

@aparrish well, it's just that they already staked out a (obviously extremely dubious and already visibly insincere) moral high ground position against OpenAI, who agreed to give Hegseth et al the "fully unlocked" anti-woke version of their competing product.

As I suspected AI Slop killed kids.

Slow clap....

kira for scale Mar 6

@jplebreton i like how even within the fairytale the narrative is now i guess "our team of compuational alchemists have created life from silicon and it's already committed mass murder"

@jplebreton Everyone should read Dario's stupid Machines of Loving grace crap fiction to learn just how shamable they are

Matt Hall Mar 6

It kinda highlights how little they understand _human rights_ when they try applying them to a really complex Magic 8-ball.

@jplebreton Anthropic's product did not kill anyone. People used it to kill. Anyone can use this to their heart intent: to build or to destroy.

@TSLST this is identical to the conservative "guns don't kill people, people kill people" horseshit that they use to defend a massive industry while mass shooting deaths pile up in the hundreds every year in this one country where people are able to do this kind of special pleading, as the rest of the world watches on rightly horrified.

@jplebreton Not even close. LLMs and IA in general are not weapons, they are levers. And quoting Anthropic when Dario Amodei got under heat for refusing to remove guardrails is plain dishonesty. US military processed data with a LLM to achieve war results. The LLM is not the culprit and the LLM is not a weapon. It tends to produce more benefits than harm though harm can be done, but Amodei's stance is to make sure human should be held responsible for what they choose to do with a hammer.

@TSLST "it's not a weapon, they just used it to pick where to drop bombs" is infantile moral reasoning, we're done here

@jplebreton They probably used a mouse and keyboard as well. Happy you understand.

@TSLST logitech doesn't market its mice and keyboards as sentient beings capable of making complex decisions, why would you think that's a remotely appropriate comparison? because you're a deeply propagandized apologist. don't bother replying.

Franchesca Mar 6

@TSLST @jplebreton guns and ai don’t kill people, but they sure as hell make it easier to kill people “at scale”.

Franchesca Mar 6

@jplebreton the only jihad worth fighting is the Butlerian one.

@jplebreton Butlerian Jihad when

@jwz 🔜

@jplebreton ah all those morons who made Clown claude #1 download last week.

While "enlightened" are busy shitting on Closed AI and copilot due to the Microslop involvement in thsoe, Captain Gemini (through its forced use through monopolistic google ecosystem)and Clown Claude (through more professional endorsement of being better at work) is winning the hearts and mind of normies and will become the duopoly in town.

Dane Deasy Mar 6

@jplebreton would it be any different with ChatGPT?

@danedeasy of course not. at the moment only anthropic is claiming any moral high ground.

@jplebreton @welshpixie And here I was grouchy at 'Anduril' and 'Palantir' getting expropriated; Anthropic's name is even worse because it's so oxymoronic.

we are now conducting war with all the moral governance of an AI-empowered ISIS.

@jplebreton I can't find sources about that appart the usual "DoD neither says yes or no". Would you have any links, I'm really interested about that topic.

@fanf42 the WSJ report on it is paywalled but various other sources, eg the Guardian, have reported on the information therein.

@fanf42 https://thisweekinworcester.com/exclusive-ai-error-girls-school-bombing/

Exclusive: AI Error Likely Led to Iran Girl's School Bombing

Pentagon investigators believe a bombing of a girls' school in Iran on Saturday likely resulted from inaccurate information provided by AI.

This Week in Worcester

@jplebreton ah, great, thank you ☺️