Allowing AI to train itself
undefined 2, 2025
Anthropic’s chief scientist Jared Kaplan warns humanity must decide by 2030 whether to let AI systems self-train, a move that could trigger an “intelligence explosion” or a loss of human control. He says recursive self-improvement is the “ultimate risk,” raising dangers of misuse, loss of oversight, and superhuman scientific capability. Kaplan predicts AIs will outperform most white-collar work within years and sees the AGI race as intense and high-stakes, urging governments to prepare and regulate as progress accelerates.
Read full article
