AI risk warnings lack evidence, experts say

Experts warn AI could become a "silicon species" or cause human extinction, but evidence for such existential risks is lacking. These claims are based on sensational narratives rather than concrete facts, according to researchers. Mhairi Aitken of Our AI Collective says AI hasn't broken free from human control, and failures stem from inadequate safeguards during testing, not rogue models. Andrew Rogoyski at the University of Surrey notes no dramatic risk shift recently, with hype driven by incidents and soundbites. Tech bosses calling for slowdowns may be seeking to avoid regulation or mask unsustainable investment costs. A real but small AI risk exists, including potential economic fallout from a bubble burst, and planning for future intelligent machines should include discussions about their rights.