Rogue AI could wipe out humanity if left unregulated, scientists warn

A rogue artificial intelligence (AI) could kill anyone if the technology isn’t properly regulated like nuclear weapons, MPs have been warned.

Researchers told the Science and Technology Committee that AI could pose an “existential threat” to humanity.

MEPs were told how the AI ​​industry has evolved into an “arms race” and that countries are competing to develop them for military and civilian uses.

They said that just as humans wiped out the dodo, the machines could wipe us out.

Michael Cohen, a doctoral student at Oxford University, said: “With superhuman AI there is a particular risk that belongs to a different class, namely . . . it could kill anyone.

“If you imagine training a dog with treats: he learns to select actions that result in him getting treats, but when the dog finds the treat cabinet he can get the treats himself without doing what we do wanted from him.”

Professor Michael Osborne, a machine learning expert, added: “I think the bleak scenario is realistic because the AI ​​is trying to bottle up what makes people special – that [quality] has led to man completely changing the face of the earth.

“Artificial systems could become just as good at geopolitically outsmarting us as they are in the simple environments of games.

“There’s some reason for hope, as we’ve been pretty good at regulating the use of nuclear weapons. AI is a comparable danger to nuclear weapons.”

Get in touch with our news team by emailing us at webnews@metro.co.uk.

For more stories like this, Visit our news page.