An AI image generator has corrected a significant blind spot, now capable of accurately depicting individuals with limb differences after initially failing to represent them.
The change came after former Australian Paralympic swimmer Jess Smith attempted to generate an image of herself using ChatGPT this summer. Smith, who is missing her left arm below the elbow, uploaded a full-length photo and specifically prompted the AI to reflect her physical characteristics. Despite repeated attempts, the AI consistently produced images of women with two arms or with prosthetic devices.
When smith inquired why the AI struggled with her request, ChatGPT responded that it lacked sufficient data to accurately create the image. “That was an crucial realisation for me that of course AI is a reflection of the world we live in today and the level of inequality and discrimination that exists,” Smith explained.
Recently, Smith revisited the platform and discovered ChatGPT could now generate an accurate depiction of a woman with one arm, mirroring her own appearance. “Oh my goodness, it worked, it’s amazing it’s finally been updated,” she told the BBC. “This is a great step forward.”
The incident highlights the inherent biases present in AI systems,which are trained on vast datasets that often underrepresent marginalized groups. The update demonstrates the potential for AI to become more inclusive as developers address these limitations and expand the diversity of their training data.
