Re:AI slays top F-16 pilot in DARPA dogfight simulation
by im3w1l on Friday August 21, 2020 @10:12AM
(#24235029)
In case of a prolonged conflict you need to consider the AI and it's programmers as a joint system. The AI by itself may be exploitable, but the programmers will adjust it to counter anti-ai strategies.
[ Parent
| on HN ]
Re:AI slays top F-16 pilot in DARPA dogfight simulation
by salawat on Saturday August 22, 2020 @05:59PM
(#24247605)
We don't program what are called AI's nowadays. We curate data, sometimes supervise the learning process, and do our best to detect when a model is bad.
If you think there's any intention or well defined predictability to these things, I've got a bridge to sell ya.