> Also. No matter how bullish you are on the potential of AI, it's horrifying and categorically unethical to use them to decide who to kill. It creates the poss

> Also. No matter how bullish you are on the potential of AI, it's horrifying and categorically unethical to use them to decide who to kill. It creates the possibility of atrocity with no accountability. No it doesn't. That was always a possibility. Atrocity with no accountability long predates AI, and the ethics of "using them to decide who to kill" are isomorphic with the ethics of using a dice roll to decide who to kill. In this case, the people you could choose to hold accountable is everybody who planned, built, or activated the machine made of a Wacky Waving Inflatable Tube Man mounting a chaingun, designed to spray bullets indiscriminately in all directions. The fact that no human trigger finger is actually on the trigger does not AT ALL defuse ethical considerations, does not exculpate anybody. We kill with machines in every military, but it's human decisions all the way down no matter how long ago a human last physically touched the munition. The lack of accountability starts to occur when a journalist embraces headlines like "AI Made Decisions About Who To Target", rather than using scarequotes and agency - "IDF-run AI system 'Made Decisions' about who to target". The IDF are the actors in this construction, the AI system is just a tool. The lack of accountability continues in the political sphere where no war crime investigation takes place, no red line is instituted, no Geopolitical Consequences arise. It finalizes in the polity when no liberal / humanist vigilante takes revenge into their own hands, no 4chah/kiwifarms investigation doxes their family, no airline pilot diverts to the Hague, we just sort of embrace treating atrocity-committers humanely because we're lazy cowards.