Tao Lin, a researcher at Anthropic, suggests that broadly superhuman AI could emerge by the end of 2027. He defines this technology as systems capable of outperforming human experts across multiple fields.
Lin estimates a 20% probability that advanced AI could eventually sideline humanity. He predicts that such developments could lead to permanent mass unemployment, forcing the population to rely on government welfare.
The researcher notes that the transition will not be instantaneous due to physical constraints. Replacing human labor requires the deployment of robotics and infrastructure, which creates a significant time lag.
To address these risks, Lin proposes slowing the pace of AI research to approximately 25% of its current speed. He argues that this would allow safety measures to keep up with rapid capability gains.
Lin emphasizes that these projections represent his personal views and do not reflect the official position of Anthropic. His forecast highlights potential long-term challenges for social safety nets.




