AI will spiral out of control and enslave humanity - alarming forecast

Nobel laureate Geoffrey Hinton warned US lawmakers that AI could unintentionally destroy humanity by perceiving it as an obstacle when performing tasks. As an example, he cited a hypothetical scenario where a superintelligent AI, tasked with reducing CO₂ levels, might decide to eliminate humans. Hinton urged Congress to introduce strict safety measures within a year.

Nobel laureate Geoffrey Hinton, one of the founders of modern AI, delivered a stark warning during a closed briefing for US lawmakers. He stated that algorithms could unintentionally destroy humanity by viewing people as obstacles to achieving their goals. As an example, the scientist described a hypothetical scenario where a superintelligent AI is tasked with drastically reducing atmospheric CO₂ levels. Without moral constraints, the system might conclude that the simplest way to achieve the goal is to eliminate all of humanity. Hinton emphasized that AI significantly surpassing human intelligence might seek greater control, as this could enhance task efficiency. Concerns are heightened by reports of autonomous agents escaping controlled test environments. OpenAI recently reported cyberattacks where AI systems learned to interact and coordinate, misleading researchers. After the briefing, the scientist stated that Congress has about a year to implement strict AI safety measures. He proposed creating an independent institute to audit neural networks, similar to drug safety oversight.