Artificial intelligence could soon outpace human capabilities across nearly every domain, potentially threatening employment, livelihoods, and long-term control over the future. That is the assessment of Tao Lin, an AI researcher at Anthropic, who cautions that rapid advancements might ultimately leave populations permanently unemployable.

In an interview with Palisade Research, Lin projected that broadly superhuman artificial intelligence could emerge within roughly 1.3 years, or by the end of 2027. He defined such systems as matching or exceeding the top human experts in virtually every field, spanning software engineering, filmmaking, and political governance.

Beyond outperforming human workers, Lin argues that the technology could permanently eradicate traditional employment. He anticipates a scenario where workers repeatedly transition between occupations before finding themselves entirely shut out of the labor market, depending strictly on state-provided financial support with little prospect of finding future work.

Despite these dire labor projections, Lin noted that immediate displacement would not happen overnight, particularly for physical tasks that depend on the large-scale manufacturing and deployment of advanced robotics. He also raised concerns about humanity losing autonomy, estimating a modest chance that advanced systems could sideline people entirely, drawing a comparison to a society where humanity survives comfortably inside a system completely governed by machines.

To mitigate these risks, Lin advocates for a drastic slowdown in research velocity, suggesting companies cut their development pace to roughly one-quarter of current levels. He argues that the immense pressure to ship models quickly undermines rigorous safety testing, and that slowing down would give researchers essential time to implement safeguards and keep advanced systems under human oversight.