Mario Soavi
983 posts

Mario Soavi
@MarioSoavi
#KM #semantic #communication #trend



I just read an article that struck a chord with me. It argued that basing AI safety on “human values” is risky because our values are not fixed, coherent, or always good. They change, conflict, and cannot be neatly separated from the rest of our minds. Trying to pin them down could make AI more dangerous. The idea that we might need better models of human motivation, or even alternatives like capability limits, makes a lot of sense to me. lesswrong.com/posts/ngqvnWGs… #AI #ArtificialIntelligence #chatGPT #OpenAI #GenAI #AGI #km #kmers











Mentre non si fermano le manifestazioni di protesta e una sanguinosa repressione, il parlamento nega le elezioni anticipate terzogiornale.it/2022/12/21/in-…

Condanna a sei anni di carcere per la vicepresidente argentina Cristina Fernández de Kirchner lnkd.in/drnH6RyS




