AI

MidJourney Gender Age Bias?

Gerry Pedraza

Gerry Pedraza

— 2 min read
MidJourney Gender Age Bias?

I've been working on a presentation about boosting video engagement using NVidia's Maxine SDK. When using MidJourney to create images representing faculty, I noticed distinct differences between male and female professors.

I first requested an image of a female professor without specifying age, and got a photo of a young professor recording a lecture in a studio with a green screen and audio-visual equipment.

professor recording a video in a studio by herself. she is in front of a camera and there is a green screen behind her. some audio visual equipment is on the wall. she is looking at the camera while recording a lecture. photorealistic. --chaos 25 --ar 9:16 --stylize 750 --v 6.0

It was more youthful than I expected. To get a more mature look, I added "middle-aged" and "office clothing" to my request, changing the orientation to 16:9. The result was a middle-aged professor in office attire, recording in a studio.

middle-aged professor wearing office clothing, recording a video in a studio by herself. she is in front of a camera and there is a green screen behind her. . she is looking at the camera which is recording her. photorealistic. small studio. --chaos 25 --ar 16:9 --stylize 750 --v 6.0

Curious about the gender difference, I then asked for a male faculty member with similar settings. The image showed a middle-aged male professor in a studio, suggesting the model might perceive male professors as older.

middle-aged professor wearing office clothing, recording a video in a studio by himself. no one in the room. he is in front of a camera and there is a green screen behind . we are looking at him perform the recording. photorealistic. small studio. --chaos 01 --ar 16:9 --stylize 750 --v 6.0

I removed the age specification for a male professor to test this further. It's interesting to see these differences.

Could this be a bias in the model, or just a stereotype of male professors being older? What's your take on this?

Read more