A new study has found that AI agents are not yet capable of conducting the open-ended research required for self-improvement. The research suggests that timelines for AI automating its own development may be overly optimistic.
log in to read full article