Scott Alexander

2026: Against SURA

In Against SURA (LessWrong, 2026-10-31), he argued that demands for safety-minded labs, employees, and countries to stop unilaterally sacrifice the backup plan of burning the lead for little gain in the chance of a global pause. He wrote that he still supports working toward a pause. In replies on X, he gave his unconditional probability of AI catastrophe as 25%.