task-completion time horizon

Large language models (LLMs) are advancing rapidly, doubling in capability every seven months, potentially enabling them to autonomously handle month-long…