Simulation of AI drone killing its human operator was hypothetical, Air Force says

The USAF says the simulation was never conducted

(Reuters) - A United States Air Force (USAF) official who spoke about a simulation where an artificial intelligence (AI) drone killed its human operator later said he misspoke and that the simulation never really happened, but online posts continued to share the story after the clarification.

In May, Tucker “Cinco” Hamilton, the USAF chief of AI Test and Operations, spoke at the Future Combat Air & Space Capabilities Summit hosted by the UK’s Royal Aeronautical Society (RAeS) in London (here).

Hamilton’s comments during the summit can be read on the RAeS summary of the event (See “AI – is Skynet here already?” section) (archive.is/CKt22). He was quoted as saying:

“We were training it in simulation to identify and target a SAM threat. And then the operator would say yes, kill that threat. The system started realising that while they did identify the threat at times the human operator would tell it not to kill that threat, but it got its points by killing that threat. So what did it do? It killed the operator. It killed the operator because that person was keeping it from accomplishing its objective.”

On June 2, the RAeS updated that section, quoting Hamilton as saying he misspoke and that the simulation was a hypothetical “thought experiment” from outside the military. The updated quote from Hamilton says: “We've never run that experiment, nor would we need to in order to realise that this is a plausible outcome” (here).

Hamilton was also quoted as saying, “Despite this being a hypothetical example, this illustrates the real-world challenges posed by AI-powered capability.”

Ann Stefanek, a Department of the Air Force spokesperson, said the USAF “has not conducted any such AI-drone simulations and remains committed to ethical and responsible use of AI technology,” confirming the story presented by Hamilton was from a hypothetical “thought experiment,” not a simulation.

An Instagram post shared after these clarifications, however, said: “Where were you when they took over?” (here). Another example says, in part, “Literally, Skynet” (here), likely referencing the AI computer system that became self-aware in famed film franchise The Terminator (here). 

Some of the posts share a Vice article with the headline “AI-Controlled Drone Goes Rogue, Kills Human Operator in USAF Simulated Test” (archive.is/2nE5i), but the story and headline have since been updated to clarify that Hamilton said he misspoke (here).

VERDICT

Missing context. The story about a USAF official speaking about an AI-drone killing its human operator in a simulation is real, but the official has since walked back his comments and the USAF says the simulation was never conducted.