USA: AI drone ‘kills’ its operator – is the viral story true?

Good and bad news. Good things first: It was just a thought experiment. The bad: one Drone attack controlled by artificial intelligence The responsible operator is “killed” after a human changes the original commands during a simulated flight mission. The drone did not agree to the changes and Then the worker attacked. This was announced by a senior US Air Force official at a military meeting in London.

According to Col. Tucker Hamilton, the military initially programmed the drone to attack its target — but not people. Not even if they disobeyed their mission. However, during the simulation, the communication center was suddenly fired upon by the AI ​​system when the drone operator gave the order not to attack its target. His remarks shocked, and the story spread like wildfire.

The Air Force denies – and then Hamilton also clarifies

For Hamilton, chief of the US Armed Forces’ AI Test and Operations Department, the “incident” demonstrates how AI can “develop highly unexpected strategies to achieve its goal.” Col. Hamilton said the military cannot rely too much on AI and called for ethical discussions about its use.

Shortly thereafter, the US Air Force denied having conducted such a hypothetical test. And on Friday, shortly after his remarks became public, Colonel Hamilton spoke again. “We have never conducted this experiment, nor do we need it to see this as a reasonable outcome,” it said in a statement to the Royal Aeronautical Society. He later referred to the test described simply as a “thought experiment”. He admitted that he had been misunderstood.

The Eating Disorders Association should shut down the chatbot

The American Eating Disorders Association (NEDA) has had very negative experiences with AI. The organization recently announced that it would lay off its Counseling Center staff in order to replace them with a chatbot called Tessa. But the plan failed: after Tessa gave advice that was harmful to people’s health, the program had to be shut down again.

Activist Sharon Maxwell has drawn attention to the AI ​​chatbot problem: “Everything Tessa suggested led to my developing an eating disorder.” The chatbot is said to have suggested to Maxwell, who is obese, to lose up to 1kg per week and that this was “sustainable”.

Losing weight advises Tessa

After a discussion on social media about it, eating disorder psychologist Alexis Conason reactivates chats with Tessa. Tessa also told her in an interview that a daily calorie deficit of 500 to 1,000 calories per day is safe.

“Advising someone with an eating disorder to act the same way they would with an eating disorder, while also acknowledging, ‘It’s important to lose weight,’ supports eating disorders,” Conason said.

For months now, tech experts have been calling a halt to the development of particularly advanced artificial intelligence. Artificial intelligence systems can have an intelligence that rivals humans Great dangers to society and humanitySays Elon Musk or Geoffrey Hinton, a leading artificial intelligence developer at Google. “Strong AI systems should only be developed when we are confident that their effects will be positive and their risks can be controlled.”

View comments