Steven Pinker Urges Pragmatic AI Safety Engineering Over Existential Doom Alarmism

Harvard psychologist Steven Pinker published an open letter in Quillette arguing that apocalyptic AI doom scenarios are exaggerated and counterproductive. He called for prioritizing pragmatic safety engineering, accountability mechanisms, and human control over speculative existential risk debates. As public and policy discussions focus heavily on existential risks, Pinker's perspective refocuses attention on immediate, concrete hazards like bioterrorism, cyberattacks, and rogue AI agents. This reframing encourages the AI industry to invest in actionable engineering safeguards rather than pursuing sweeping development halts or theoretical doomsday panics. While Pinker acknowledged underestimating both the rapid capabilities of large language models and the recklessness of tech companies in release races, he argued that thought experiments like the 'paperclip maximizer' conflate raw intelligence with a drive for dominance. In contrast, blogger Scott Alexander estimated a 20% chance of AI-induced human extinction and advocated for slowing development via international treaties.