Simply put, because demand for most of the things that humans produce is much more elastic than we recognize today; and as long as humans are complementary to the production process, it won’t be rare for efficiency gains to get swallowed up by demand growth. This is the famous Jevons paradox, the tendency for the more efficient use of a resource to increase total consumption of that resource, rather than decrease it. Energy is the classic Jevons case: we find over and over again that as energy becomes more efficient to produce, people respond not by consuming the same amount of it, but by increasing their consumption—such that overall energy use tends to rise. (Thus the “paradox” part: energy efficiency increases energy consumption!)
Take software. Software simply means “the things that a computer can do”: and because software is so broad and so capable, we should expect that it’s energy-like and that there’s an enormous latent demand for more software in the world. For that reason we’ve found that every increase in the efficiency of software programming—the move from lower-level to higher-level languages, the emergence of frameworks and libraries, the endless move away from bare metal as compute becomes ever-cheaper and more abundant—has resulted in a dramatic increase in demand for software engineering labor.
All of this makes me suspect that, as long as we are in the cyborg era of human-AI complementarity, we should be quite optimistic about human labor. The world is governed by bottlenecks; as long as those bottlenecks are real, there will be complementarity between humans and AI; and if the human-AI combination can make human labor vastly more productive, we should expect that to be a very good thing: for consumers of course, who will benefit from an enormous consumer surplus, but also for workers.