1 in 20 doesn't sound great. But you have to mediate that with

- everyone and their mother are doing a "generative ai program" right now, a lot of times just using the label to try to get their project funded, ai being an afterthought

- if the 1 out of 20 projects is game-changing, then you could argue right now people should actually be willing to spend even more on the opportunity, maybe the number should actually be 1 in 100. (The VC model is about having big success 1 in 10 times.)

- studies of ongoing business activities are inherently methodologically limited by the data available; I don't have a ton of confidence that these researchers' numbers are authoritative -- it's inherently impossible to truly report on internal R&D spend especially a private companies without inside information, and if you have the inside information you likely don't have the full picture.