Managing Our Agents

Peter Bihr, in Do we all want to be middle-managers for AI assistants?, writes:

These AI assistants are like putting an intern in charge of your life. Do you really want to middle-manage all of this, focus on quality control and all, rather than just do the thing yourself? The doing, that friction that it brings, that is the life, that is the work!

I mean, who wants to be a middle-manager for their own life?

An echo of a similar sentiment from China that I linked to yesterday.

Peter’s reference to his father’s generation reminded me that there was a point in my father’s career as a research scientist where he was promoted to middle-management. 

At the Canada Centre for Inland Waters there were sections and divisions, and at some point, in his late 40s if memory serves, he was promoted to become a section head. He hated it, eventually gave up the position, and returned to being a full-time researcher for the rest of his career.

The imprint on me of this, which I recall being a frequent topic of dinner conversation at home, was that there are doers and there are people who manage the doers, and that there is a nobility in being a doer, and a strong crass, demoralizing quality to being a manager.

Peter asks “I mean, who wants to be a middle-manager for their own life?”, and I think he’s speaking directly to that schism: there is joy, accomplishment, self-knowledge, in the friction of doing. On the other hand, once we get beyond the thrill of bossing agents around and having them do our bidding (and becoming their managers), the demoralizing realization that, in Cory Doctorow terms, the horses are riding us.