«Godfather of AI» warns of danger to humanity

Nicknamed the “Godfather of AI”, Geoffrey Hinton reiterated his concerns about the “tech bros” approach to artificial intelligence.

Having already warned that there is a 10% to 20% chance that AI will cause the extinction of humanity, he believes that AI companies are adopting the wrong strategy, fueled by fierce competition around the technology.

According to Hinton, the future of AI is getting «increasingly scary», which is hardly reassuring.

At a recent conference in Las Vegas, Hinton said that industry giants were taking an inappropriate approach to curbing both the desire and development of AI’s ability «to stay alive», but also its ambition «to get more control» in the face of human unpreparedness for it.

A recent disturbing episode occurred during tests carried out by Anthropic on its AI model, Claude Opus 4. In a fictitious scenario, the model behaved in a worrying way, threatening an engineer with revealing his extramarital affair in order to avoid deactivation. This type of reaction appeared in almost 85% of the trials, involving not only blackmail, but also other worrying actions, such as attempting to clandestinely transfer one’s own data, lock out users, or contact the authorities or press.