2 Comments

Very interesting! It seems to me that this concept of reflexivity can be applied to graphics cards and machine learning as well. When graphics cards first became a thing in the 90's nobody envisioned the massive market for machine learning workloads that would arise decades later. That market was became available partially because of the availability of fast graphics chips to do massively parallel computations.

Expand full comment

great post and highly relevant. It's akin to Nassim Taleb's "Second-order thinking" --> the major consequences of any new technology (AI chips) or socioeconomic event (Ukraine invasion) are largely unpredictable as they are masked by non-linear relationships.

Expand full comment