‘The People Building AI Earnestly Believe That It Could Kill Us All’: Anthropic Researcher Quits Dramatically

It’s very easy to imagine war machines with a shoot on sight algorithm. It doesn’t have to be AGI for it to be good at hinting people down.