If there’s one concept educators can no longer afford to ignore, it’s systems thinking. It’s not new. But it’s newly urgent.

Chain-of-thought monitorability could improve generative AI safety by assessing how models come to their conclusions and ...