Self-defeating example: betting on red in roulette is a bad idea. Having your friend bet on red is a better but still bad idea. Having a stranger bet on red, and buy you a drink if he wins... It's free money.
Applied example: starting an electric car company (at least looked like) a bad idea.... trying 1,000 different approaches to a light bulb, especially after 500 failures (at least looked like) a bad idea... etc.
I worry about a Blockchain winter, then again AI has made fantastic progress despite the AI winters[0].
Would fusion as a power generating technology be further along or further behind if it had followed the AI model (progress -> hype -> winter -> progress)?