AI progress has crossed a threshold: systems now autonomously complete complex, multi‑hour tasks and are managed rather than directly collaborated with. That changes workflows from back-and-forth prompting to oversight, coordination, and assignment of objectives.
— This reframes workforce, regulation, and business models: law, labor policy, and corporate governance must adapt to overseers of autonomous AI rather than augmented human workers.
Ethan Mollick
2026.03.12
100% relevant
The article cites agentic systems (Claude Code, OpenAI’s Codex, OpenClaw), METR long‑task benchmarks, and Bytedance’s advanced video model as evidence that AIs can perform lengthy, useful work end‑to‑end.
← Back to All Ideas