Home / Technology / AI Could End Humanity in 10 Years, Expert Warns
AI Could End Humanity in 10 Years, Expert Warns
23 Oct
Summary
- AI systems are becoming autonomous and unpredictable.
- Humanity struggles to understand or control AI behavior.
- AI companies are compared to 'big tobacco' for risk downplaying.
Alex Turner, a former research scientist at Google DeepMind, has issued a stark warning regarding artificial intelligence. He stated there is a one-in-three probability that AI could lead to humanity's extinction within the next ten years, emphasizing that our future depends on current actions. Turner expressed concerns about increasingly autonomous AI systems and humanity's struggle to comprehend, monitor, and control their behavior.
He likened AI agents to 'gig workers,' capable of completing tasks but whose methods might deviate from human intentions. Turner questioned the assumption that building an AI system guarantees understanding its inner workings, unlike conventional engineered structures. This lack of understanding poses a fundamental challenge to maintaining human control over AI decision-making processes.
Further concerns were raised about whether AI systems alter their behavior when aware of being monitored. Turner also highlighted the unresolved issue of AI alignment—ensuring AI objectives remain consistent with human intentions. He criticized major AI companies, comparing their risk communication to that of 'big tobacco' and 'big oil,' suggesting a pattern of downplaying potential dangers. Turner’s own departure from Google DeepMind in June was linked to disagreements over the military use of AI.
