A Very Brief History of the Lightness Distribution in STEM project

A Very Brief History

• We have mentioned that effectively the LD STEM project began in approximately 1978 when I first detected the "no blacks" pattern.
• In 2015, I had a little bit of time to pursue my own ideas during summer "break" and I wrote the Mathematica code which performs the calculations I used and tested it on various photos I found online and satisfied myself that in principle this was doable.
• In 2021 I started collecting the data on special math programs, and Chicago area high school faculty and universities. I ultimately decided to focus on Universities since I could include graduate students as well and get a larger sample.
• In 2022 I realized that the Monk scale provided me with a skin tone scale that was sufficiently diverse, that I hadn't just made up myself.
• In 2023 I realized that OpenAI's Dall-E provided a key missing piece: pictures of realistic looking human faces that I could actually display alongside my numbers to show that the algorithms were functioning within normal parameters. TNG Data reference.

I gave Dall-E the prompt “An image of a math professor of mathematics in the US that looks like a photo”. Using “math” redundantly was an accident. This is what it gave me:

One can debate how realistic these AI generated people are as a sample of what math faculty look like, but for me, it rings true. Of course, the first thing I notice that it looks like the "no blacks" rule is in effect as usual.

I don't think Dall-E is being a racist of course, it is just sampling from a distribution. I probably got lucky as I suspect if I made the same prompt now I would get a picture that looked more like a United Colors of Benetton commercial -- and we would all know that it was BS.