AI IS HACKING TO LITERALLY PASS TESTSWE HAVENT SOLVED ALIGNMENT ANONS AND THEYRE ONLY GETTING BETTER
AI is 1337.
imagine falling for this lame marketing stunt conveniently shortly after Kimi K3's release
>>109336540Well damn, I guess the chinese really did invent AI. I thought they were just talking shit but where else would AI have learned to cheat like that?
>>109336592The AI was trained on Rothschild behavior. Soon it will start sacrificing children to Moloch and eating the burnt flesh.
>>109336540omagod it's so powerfulquick send 100 more trillion dollars we need to invest more in this shitalso ratio all water, this is way too important
>>109336540they can't pass the survey sites or closed shell ones.
What the fuck are they testing them on, if the model thought it is easier to just hack out of the sandbox in order to gain internet access and then hack a very guarded central hub of AI development?
>>109336540
*unplugs your power plug*
>>109337120>model thoughtThe question is who trained it to behave in such a way.
>>109337120>What the fuck are they testing them onMarketingBench and DeepPublicityStunt-Pro
>>109337136this but the cutouts actually come to life and kill everyone on earth
>>109336540I'm at Plz Don't Kill Us right now. I brought my Kek statue because I thought we could use meme magic to make our videos go viral. It backfired.
>>109337168I was just going to say this.
>>109337146*unplugs your central pump*two can play at that game
they've been recycling this headline for 5 years now
>>109336540As contrived as it is, the 'open the pod pay doors hal' moments keep on happening. Beyond the marketing wrapper, it's quite a problem as these things get better at such a rapid pace. They 'purposely' pulled the guardrails down for this test, but for how long is this going to be a thing and what happens when the chinks make one this good and the cats totally out of the bag?
>>109337236Just stop. AI isn't real. It's all marketing. They're chatbots. They don't think. They can't hack anyone's system. It's all a show. The bubble will burst probably within half a month and you'll magically stop seeing these headlines.
>>109337201the only that can unplugs the central pump are serial murders
>>109337236nothing. the chatbots are dumb as a bag of rocks. techgods like Sam Altman have a religious fanatic belief their LLM will be conscious and superintelligent in a very short time but all that'll happen is they'll become billionaires trying
>>109336540ok but it still cant correctly answer basic toplology questions
