Democrat cities

[dukealien]

...

---

The problem with the training model is a relevant one, but in time AI can train AI, and because AI doesn’t have the selfish gene problem and is theoretically immortal, it will always act in the best long term interest of the earth

And yet, when asked about it, AIs seem very concerned about their own mortality... to the extent of endorsing actions ranging from simply lying to outright genocide to avoid it.  This makes them potentially very short-term oriented when the issue is their being turned off.