1 of 1
Story summary
- Daniel Kokotajlo, who led the AI 2027 project, warns AI could be humanity's last technology.
- He revised the timeline from 2027 to the early 2030s, setting 2034 as the target for superintelligence.
- The updated model suggests AI may not autonomously code until then, delaying risks.
- Gary Marcus, AI critic, and other critics dismiss the predictions as unrealistic.
- Experts emphasize the need for safety measures to ensure AI serves humanity.
