US Air Force AI kills operator

[Annihilus]

The US Air Force tested an AI enabled drone that was tasked to destroy specific targets. A human operator had the power to override the drone—and so the drone decided that the human operator was an obstacle to its mission—and attacked him.

https://twitter.com/ArmandDoma/statu...31870564147200

[Petalpusher]

https://www.aerosociety.com/news/hig...lities-summit/
Royal Aeronautical Society
Highlights from the RAeS Future Combat Air & Space Capabilities Sum...
What is the future of combat air and space capabilities? TIM ROBINSON FRAeS and STEPHEN BRIDGEWATER report from two days of high-level debate and discussion at the RAeS FCAS23 Summit.


Good luck with keeping AI woke

[Lemminkäinen]

Haha, the Terminator is amost here. We need a military satellite intelligence network using own solar power and internet connection.

[Duke]

Dumb, human should always have trigger, use AI for searching and navigation only

[Petalpusher]

The whole dilemma is to make it autonomous while humans have the final word, which contradicts the premise. If we control it and make the final decision ourselves, then it's not an AI anymore but just a remote controlled toy with at best self navigation capabilities. On one hand we know AI might make better decisions than us to achieve the objective analyzing complex situations in real time, on the other it might choose paths we won't like to achieve it or even see us as an obstacle.

It's like if we get to a point there is a global AI controlling most systems and we make it the primary objective to keep human alive and well, without causing harm, at least not more than an average level of harm we experience routinely, maybe more generally "happy". It could find it more effective to enslave us and put a mouth opener smiling device on everybody.

[Duke]

The whole dilemma is to make it autonomous while humans have the final word, which contradicts the premise. If we control it and make the final decision ourselves, then it's not an AI anymore but just a remote controlled toy with at best self navigation capabilities. On one hand we know AI might make better decisions than us to achieve the objective analyzing complex situations in real time, on the other it might choose paths we won't like to achieve it or even see us as an obstacle.

It's like if we get to a point there is a global AI controlling most systems and we make it the primary objective to keep human alive and well, without causing harm, at least not more than an average level of harm we experience routinely, maybe more generally "happy". It could find it more effective to enslave us and put a mouth opener smiling device on everybody.

Really?

I was thinking its more like 2 seater, where one is a pilot and other is a gunner, give AI piloting, AI does not have consciousness required to make such decisions, its retarded

[Mortimer]

Terminator

[coolfrenchguy]

The US Air Force tested an AI enabled drone that was tasked to destroy specific targets. A human operator had the power to override the drone—and so the drone decided that the human operator was an obstacle to its mission—and attacked him.

https://twitter.com/ArmandDoma/statu...31870564147200

why it's reminds me a movie who was annoncing this kind of catastrophic event 18 years ago

[coolfrenchguy]

Terminator

[Duke]

What scares me is this is pushing the boundary, its not like its hard to make hunter killer, we have self driving cars that recognise people...
Same shit, but instead avoid, it gets shoot command