There's been lots of talk about AI safety here lately. Ilya Sutskever recently started a company called SSI - Safe SuperIntelligence. And the OpenAI CTO has warned repeatedly, but vaguely, about AI dangers, all as other Open AI employees have quit while citing the company's excessively relaxed approach to safety.
But many of us here have been frustrated by a lack of specificity of what the danger is. But I think I just accidentally stumbled upon an AI danger so scary that we're not allowed to discuss it . I just had a post deleted by the moderator because I described in some detail an idea inspired by one of Depeche Mode's biggest hits, and how AI might be used to implement the concept in the song, and what the ramifications might be.
[link] [comments]