False alternatives and the economics of AI
Here's a question du jour: When software companies lay people off, are they doing it because of AI or because they over-hired?
Other explanations are obviously possible. Even if we only think about these alternatives, though, we need to keep some important points in mind:
- These alternatives often describe the same thing at different levels of explanation. Over-hiring doesn't usually mean you literally stockpile people on the Hooli roof. Instead, you overstaff teams, try experimental projects, and so on. AI is very good at helping eight people do what nine people are doing right now; if the project was overstaffed, that might be seven (or three, or one, or zero). Overhiring means there's more to cut with AI; they are both partial descriptions of the same phenomenon.
- Accounting for the mechanics of getting work done is different from negotiating the political meaning of layoffs. The question of how much headcount is needed to do X is different from the question of how a layoff is going to be made intelligible to staff, shareholders, and the public. In 2022, it was a lot more popular to say "we value investment and engineering" than "money is cheap, why not." Here as elsewhere, meaning is negotiated after the fact--in part because different accounts are accurate at different levels of explanation.