"Air Force AI Drone Goes Rogue & 'Kills' Its Human Operator in Constructed Simulated Test !" Benjamin Fulford said over 16 Japanese scientists working on military offensive robot were killed recently by their research bot. Looks like Skynet is pending. https://www.rumormillnews.com/cgi-bin/forum.cgi?read=223633
They did say it was a constructed simulated test. Meaning, if it had been real, the operator would have been killed. The powers of discrimination of a lethal robot are not to be trusted.
The Japanese incident is based on Fulford's credibility, which is often correct but also sometimes hit and miss. He is quite reliable on Japanese matters because he lives in Japan and has written books in Japanese on economics and politics (he is originally Canadian and his father was a Canadian diplomat).
Yes, it's true that an AI will simply do anything to attain its goal, that's all it's programmed to do. If you've ever seen a tool-assisted run on a video game (TAS) you'll see some of the strangest stuff. An AI is programmed to do a speed run, or to achieve the highest possible score, and the result is that, in time, the AI finds nearly every bug and glitch to attain its goal.
However, with a lethal system it's pretty easy to put in a kill switch and simply change its directive.
This is just more nonsense from our bullshit news media I think.
https://www.rumormillnews.com/cgi-bin/forum.cgi?read=223633
https://twitter.com/TaskandPurpose/status/1664388113383280640?ref_src=patrick.net