Post #1885479

2026-04-30 06:15 UTC

@jackemled @freddy if anything, they’re exaggerating how important the AI was to this endeavor: the fundamental shape of the flaw was identified by a human in design review and they used the AI to “scale the research,” which is pretty vague

Replies (1)

@peritia@alovely.space 2026-04-30 06:36

@0xabad1dea @jackemled @freddy I am pretty sure some people forget that AI does not think. It is still just a proof of the "Infinite monkey theorem" In the end it is just typing random words (actually it is tokens but eh) and then uses statistics to predict what comes often after. Just like the next word prediction in phone keyboards Just because it new the next word i wanted to type was "kangaroo" doesnt make start glazing it more than the donut from the donut store. Not to mention: These ai vulnerability scanners will get worse by the day. Ai scans code by people using ai to write code deems vulnerablities as normal and starts being shit. "yes but people will still check the code" Then why use AI in the first place?!?

Open ##1885480