> AI still has a long way to go before the Air Force pilots would be ready to hand over the stick to an artificial intelligence during combat, DARPA officials s

> AI still has a long way to go before the Air Force pilots would be ready to hand over the stick to an artificial intelligence during combat, DARPA officials said. It seems to me that rival states who also develop AI-enhanced systems for their respective fleets may do a different cost-benefit analysis and throw caution to the wind on that point. A foreign fighter craft such as a MIG, traditionally with lower maneuverability when pitted against the likes of an F-22 (and similarly well-piloted), can eke out a significant equalizing advantage if flown by an advanced AI system; especially under the duress of protracted skirmishes. And let's not forget the industrial cost of building AI-powered jets at scale vs. training highly-skilled human pilots en masse. That's already showing significant ROI in the form of Predator drones. I think combat-ready AI fighter jets in the US will progress along a development roadmap that matches their growing need during acts of deterrence, rather than during full-scale war. But when multiple states have sufficient AI capability to go tit-for-tat without loss of life, then that would signal an ability to conduct more strategic "stress-tests" that carry with them a significant possibility of escalation.