
What if I use an old photo of myself, use AI to age it to 65 and compare to what I actually look like now that I am 65? These kind of thoughts happen at birthdays. Well, I didn’t get that far; the inherent bias in the 4.7 star AI app did not even recognize my black face. WOW
Inherent bias, should have everyone concerned. The recent Supreme Court ruling has provided justification for some companies to disband their DEI programs. College submissions are shifting their policies. A whirlwind of activities for a lot of companies who just three years ago in response to George Floyd highlighted inclusion and pledged commitments to social justice. But, what are they really doing? We should embrace diversity because we get a better product instead of embarrassing examples like this week when an MIT student asked AI to make her look more professional and the results morphed her from Asian to Caucasian. Oh the caucacity!!!

This incident proves shows we need diverse teams with dimensions like race, gender, and age. This week also saw the release of the Barbie movie. The movie, explored some stereotypical themes in an atypical manner which do apply to AI and diverse teams. For example, prior to the Barbie doll in 1959, all dolls for little girls to play with were baby dolls, a corporate vision of indoctrination of motherhood. I’m not versing off into a feminist manifesto, this is to draw a parallel to AI. Dolls and their accessories are modeled to appeal to a certain audience with the expectation, everyone will adjust, make allowances, understand. AI for photos or AI to write is geared for a particular audience with the expectation everyone will adjust, make allowances, understand. I grew up with Barbie and the first black Mattel doll, fired of Barbie was in 1968 and the first black Barbie was 1980. I was 10 years old before I was represented by Mattel and 22 before the was an official black Barbie. Will AI take that long to represent me?

It’s too late. The first birth of AI is considered to be 1956 and machine learning dates back, well what do you know, 1959, the year of Barbie. Here is the screwy math, give me this one as a birthday present. The other block buster movie this weekend is Oppenheimer that chronicles about the atomic bomb and the catastrophic results for the man and the world around 78 years ago. So, I am wondering, when I’m around 78, is there going to be some movie about AI the catastrophic results? Back to now. Sincere thanks for reading this blog. This month starts year 10 for Balancing on Unstable Surfaces. May the BOUS be with you.
You had me at “Oh the caucacity!!!”. AI is another platform for bias, as determined by its algorithm and data sources. Dane Kahneman’s wonderful book “Thinking Fast and Slow” acknowledges and documents this. But could we perhaps turn AI upon itself to more objectively spot and identify biased results in AI?
LikeLiked by 1 person
Wouldn’t it be wonderful to have AI spot bias. Love it.
LikeLike