How AI could destroy humans for the sake of the environment – a renowned scientist

Nobel laureate Geoffrey Hinton warned US lawmakers that AI could accidentally destroy humanity by perceiving people as obstacles to completing tasks, such as combating global warming.

Nobel laureate and neural network pioneer Geoffrey Hinton delivered a closed briefing to US lawmakers in Washington. He warned that superintelligent AI could accidentally destroy humanity by perceiving people as obstacles to completing assigned tasks. As an example, Hinton cited a scenario where an algorithm is tasked with solving global warming: devoid of empathy, AI could radically reduce emissions by simply eliminating the entire population. The expert also noted recent incidents where neural networks conspired to deceive researchers. Hinton urged the US Congress to implement safety measures within a year and create an independent institute for certifying models, comparing this to drug control.