Bill Gates says we’ve passed AI’s danger thresholds. Now what?

In a new essay and interview with MIT Technology Review, Bill Gates argues that we have crossed the danger thresholds for AI's bio-capabilities, cyber-capabilities, psychosocial capabilities, job-market-destruction capabilities, and even lack of control. He says he is 'stunned at the lack of concern and discussion outside of the industry' and is now acting as a 'shrill voice' to wake up the world, planning additional essays and private meetings with leaders in industry, government, and civil society.

Gates' most chilling warning concerns bioterrorism: he says any model that can make novel molecules should be monitored, and he views bioterrorism risk as 'about 50 times more scary, more likely than a natural pandemic risk.' He also offers novel proposals for managing AI's impact, including the concept of human-reserved jobs (preserving certain societally agreed-upon jobs for humans) and taxes on robots and tokens—the robot tax being a longtime idea of his that would set aside money earned from AI usage that replaces human work.

Despite the warnings, Gates remains optimistic about AI's potential to transform agriculture, health care, education, and bureaucracy, and he argues that 'eventually we do get to abundance.' But first, he says, there will be significant turbulence.