One finding of an extensive theoretical and empirical economics literature on technology races is that firms that trail the current leader might tend to innovate more (Reinganum 1983; Lerner 1997). My question is about whether the firm-level payoffs to innovation in the ~AI industry might cause similar patterns.[1] The next paragraph very briefly summarizes the existing literature in order to clarify my question.
In the disk drive industry, the firm that managed to develop the disk drive with the highest storage density at the time earned a commercial advantage—it could take significant market share away from its competitors. This created an obvious incentive to innovate, for all firms. Somewhat counterintuitively, some theoretical models (e.g., Reinganum linked above) imply that this incentive is most strongly felt by firms trailing the current leader in disk drive density, rather than the leader itself. This is because, if the current leader invents a superior disk drive, it mostly retains its current market share. But if a trailing firm invents a superior disk drive, its payoff is much larger; it can seize market share from the current leader and other trailing firms. As a result, most innovation comes from trailing firms, rather than the current leader.
Why does this matter?
I’m not sure. I mostly ask out of curiosity. But if you expect ‘challenger’ firms to drive innovation more than ‘leader’ firms in either hardware or software for AI, this should adjust your beliefs towards a neck-and-neck race for AI. That is, as opposed to a world where one firm is leaps and bounds ahead of the rest, and has the time and money to pay an alignment tax.
- ^
The ‘AI industry’ is a deliberately vague term. I’m curious as to whether we observe this specific race dynamic in any of semiconductors, LLMs, etc.
Having done some of this modelling myself, I think it’s difficult to pin down the exact outcome of a particular race. Some empirical evidence suggests that winning a patent race leads to more follow-on innovation, while other models, including those fitted to data, suggest that laggards are often more innovative. However, models also suggest that laggards who are quite far behind tend to give up racing entirely.
My tentative conclusion is that the finding you highlight is plausible enough such that I’d consider small gaps in innovativeness to ~= neck-and-neck races, but larger gaps to produce a monopoly-like situation for the race leader. Determining where precisely this cutoff, of course, is difficult.