Reminds me of the story of WWII (I believe) US fighter plane cockpit design, in which the Army commissioned a study to find the average size for a male in all t

Reminds me of the story of WWII (I believe) US fighter plane cockpit design, in which the Army commissioned a study to find the average size for a male in all their components (average torso, average thigh length, average forearm length, etc.) so they could build the best plane. Turned out that it fit no one at all, because no one has more than 3 of these at an average size. As these cockpits fit no one, casualties from crashes due to inability to handle the plane became significant. The answer was to make everything adjustable, of course. Ref: [https://www.thestar.com/news/insight/2016/01/16/when-us-air-force-discovered-the-flaw-of-averages.html https://www.thestar.com/news/insight/2016/01/16/when-us-air-...] So statistical probability of details does not work in favor of the bigger picture. Pictures of people made up of average components do not look average, they look weird. At least the ones I've seen. And, the big takeaway is that GPT doesn't have knowledge yet, but rather a lot of words.