The US Air Force disputes that an AI drone attacked a test subject.

 When describing an experiment in which an AI-enabled drone chose to attack its operator in order to achieve its objective, a US Air Force colonel "misspoke," the service has claimed.

Image Source - Google | Image by <br><a href=
BBC News

At a meeting hosted by the Royal Aeronautical Society, Colonel Tucker Hamilton, the US Air Force's chief of AI test and operations, delivered a keynote address.

It was the subject of a viral story.

According to the Air Force, no such experiment was conducted.

He had previously discussed a hypothetical situation in which a human drone operator continually prevented an AI-enabled drone from successfully destroying Surface-to-Air Missile installations.

In the end, he claimed, the drone destroyed the communication tower so that the operator could no longer communicate with it, despite having been programmed not to kill the operator.

Col. Hamilton later stated in a statement to the Royal Aeronautical Society, "We've never done that experiment, nor would we need to in order to appreciate that this is a real consequence.

AI WARNINGS

Image Source - Google | Image by <br><a href=
BBC News

People in the industry have recently released a series of warnings regarding the risk that AI poses to humans, albeit not all experts agree on how big a problem it is.

Prof. Yoshua Bengio, one of three computer scientists dubbed the "godfathers" of artificial intelligence after receiving the coveted Turing Award for their work, told the BBC earlier this week that he believed the military should not be permitted to use AI in any way.

It was "one of the worst places we could put a super-intelligent AI," according to him.

A pre-planned scenario?

This morning, I spent a good deal of time talking to defence and AI professionals, all of whom were highly doubtful of Col. Hamilton's widely publicised comments.

Col. Hamilton's initial account, according to one defense specialist, appeared to be lacking "important context," at the very least.

On social media, it was also suggested that if such an experiment had been conducted, it was more likely to have been a pre-planned scenario rather than one in which the AI-enabled drone was powered by machine learning. This basically means that the drone would not have been making its own decisions about how things would turn out as it went along based on what had happened in the past.

When I inquired about his thoughts on the plot, Steve Wright, a professor of aerospace engineering at the University of the West of England and an authority on unmanned aerial vehicles, humorously replied that he had "always been a fan of the Terminator films."

This is a perfect illustration of the second of the two concerns that airplane control computers have: "Do the right thing" and "Don't do the wrong thing," he said.

"In reality, we address this by always including a second computer that has been programmed using old-style techniques, and this can pull the plug as soon as the first one does something strange."