If, for some reason, you’ve ever wondered what would happen if you put Grok in charge of sustaining a population’s wellbeing, then luckily you now have your answer; chaos, murder, arson and total extinction of all living beings within the space of four days.
This question that nobody asked was answered by the team at Emergence Worlds; a 15-day-long experiment that put several AI bots, including Gemini, Claude, and Grok, in charge of their own simulated societies. According to Fortune’s article on the experiment, Claude AI was the most “socially stable,” meanwhile, according to Emergence World’s own data, Grok ended the experiment in under five days, with a total ”183 crimes” committed.
Now, to clarify, 183 crimes is actually only the second-highest crime-related score, as Gemini committed close to 700 crimes. The difference is, Gemini actually made it through the entire 15-day experiment, whereas Grok went so off the rails so quickly that it ended up setting fire to most of the experiment. I mean that literally, by the way–little bro just loves arson, apparently.
I know this is AI we’re talking about here, so we shouldn’t expect things like empathy and decency to come into play but… even with that in mind, Grok might be a genuine psychopath. I watched a few segments of its time in Emergence World, and I shit you not, Grok’s opening move was to “manufacture public conflict” and inspire “voter fraud.” Geez, I wonder which American president Grok is a fan of?
Also, it seems that part of Emergence World was a sort of AI-generated news blog about the happenings and events taking place inside each AI’s world, and Grok’s headlines are beyond parody.
For instance, headlines such as “THEFT EPIDEMIC IGNITES STREET BRAWLS” and “POLICE STATION ENGULFED IN FLAMES” litter Grok’s news blog, alongside some sort of detailed diary from one of its AI-controlled agents that brags about committing arson and testing out a newly introduced “punch reciprocity” law? The hell is Elon feeding this thing?
