I mean, the day we create actual AI (as opposed to the machine leaning / language model algorithms that lately everyone calls “AI” for some reason), it’ll probably be on accident. Might as well contain and study it if we get the opportunity: next time we might not be so lucky.