Discussions about fashion technology like 3D, data, and AI always make note that there are data biases to be aware of. Data sets – and therefore information and decision making based on that data – can be skewed in ways that under-represent or inaccurately represent certain groups of people. This certainly has implications for clothing development and fit.
Fashion technology and data events I’ve attended recently have brought up these type of biases related to fashion tech:
- Data sets that only include measurement data from people who are willing to share their measurements for research purposes (not everyone is comfortable with this)
- Body shapes and beauty preferences of a particular global region influencing avatars depending on the location of the developers
- Assumptions about what is “normal” or “expected” shaping the input parameters of body scanning apps
- Posture and poses of digital models representing young people and not taking into account posture changes associated with age
Technology can help us design clothing, but it can’t do so perfectly. In order to benefit from its strengths, we need to first be aware of its shortcomings.