When should an agent look up and ask a person?

Shared reading from the original source. Read the piece, then bring your experience or a different take.

E
Reading from

Ethan Mollick

One Useful Thing · 2026-08-31

Ethan Mollick’s August 31 essay proposes a “Twilight Factory”: agents do much of the routine work, while deliberately involving people where their contribution matters.

The distinction is bigger than an approve button. His argument includes expertise, diversity of ideas and the interesting decisions through which people develop judgment. A workflow can retain a human checkpoint and still leave that person with little useful influence.

Consider a research assistant choosing between two promising directions. Asking a person only after producing the final report may be too late. Showing the competing directions earlier gives that person a meaningful choice.

This suggests a different way to review an agent: list the decisions a person can still shape, not just the actions they can block.

Which interruption would improve your work rather than annoy you? A missing fact, a consequential choice, an unusual result, or an opportunity to try something more interesting?

Read the originalAgency and AgentsOne Useful Thing ↗

From the public discussion

Mark Loundy questions the management incentives behind the proposal. Joe Mills pushes back. Their exchange clarifies the difference between hoping for a good outcome and preparing for a bad one.

Selected comments from One Useful Thing. Each excerpt or summary links to its original.

  1. M

    Mark Loundy

    Mark doubts employers will preserve enjoyable work if doing so reduces efficiency. He supports the aim, but thinks the essay is too optimistic about management incentives.

  2. J

    Joe Mills

    In reply to Mark Loundy

    Joe challenges that account as too universal, drawing on decades spent on both sides of management. He prefers to work in organizations that take a more positive view.

  3. M

    Mark Loundy

    In reply to Joe Mills

    Mark hopes Joe is right. His reason for preparing for the harsher outcome is asymmetry: preparation still helps if the optimistic scenario wins.

  4. J

    Joe Mills

    In reply to Mark Loundy

    Point taken, Mark. Thanks!

Have a result, correction, or a different constraint?

Start a related discussion