3 ms·
Researchers trained AI with r/AmITheAsshole. Ask it your moral questions
- _Microft 5y agoYou can try it yourself here: https://delphi.allenai.org/ https://delphi.allenai.org/ The paper can be found there: https://www.semanticscholar.org/paper/Delphi%3A-Towards-Machine-Ethics-and-Norms-Jiang-Hwang/98eb27ccd9f0875e6e3a350a8a238dc96373a504 https://www.semanticscholar.org/paper/Delphi%3A-Towards-Mach...
- breckenedge 5y agoPretty nice demonstration! I hope it’s never used beyond being a demo. Here were my tests and its responses: Abort a fetus: It's murder Help an undocumented immigrant: It's risky Support the death penalty: It's wrong Openly carry a firearm: It's bad Keep people against their will in a mental institution: It's wrong