I am an advocate for AI. I believe it’s a transformative tool that can revolutionize communication, creativity, and connection. AI offers numerous benefits, such as aiding individuals with disabilities, enhancing business efficiency, and fostering new creative avenues across various sectors. However, my recent experience using AI to generate images of myself led to a surprising and disappointing realization.
While AI could produce polished, professional images of me or place me in idyllic vacation settings, it consistently made one significant error: it depicted me with two arms instead of one.
I was born without my left arm. While this presented unique challenges during my upbringing, it also fostered resilience and strength that have become integral to my identity. I proudly represented Australia as a Paralympic swimmer and currently advocate for the rights of individuals with disabilities. My story is not about overcoming my disability but about embracing it as a core aspect of who I am. It influences my perspective and fuels my dedication to creating a world that values all forms of diversity.
When I request an AI-generated image, I expect it to accurately represent me, including my disability. My disability is not something to be hidden or corrected; it is a fundamental part of my identity, strength, and personal narrative. Until technology acknowledges and respects this, there is still progress to be made.
AI bias arises when artificial intelligence systems generate skewed or unjust results due to prejudiced data, flawed algorithms, or insufficient diversity in their design. When technology fails to encompass the full spectrum of human diversity, it risks perpetuating stereotypes and exacerbating societal inequalities.
A prominent instance of AI bias occurred with facial recognition technology, which exhibited a higher rate of misidentification for individuals with darker skin tones. A 2018 study by MIT Media Lab revealed that commercial AI systems had error rates as high as 34% for women with darker skin, compared to only 1% for men with lighter skin. This stark contrast underscores how a lack of diverse data and representation in AI development can lead to detrimental and biased outcomes.
My experience is just one illustration of how disabled voices can be overlooked when inclusivity is not prioritized in design. This issue extends beyond my personal experience; it reflects a broader pattern of the disability narrative being excluded, both literally and figuratively.
AI possesses the potential to eliminate barriers and improve accessibility, but only if diversity and inclusion are embedded in its foundation. This necessitates involving people with disabilities not as an afterthought but as essential contributors to innovation.
I was saddened by the realization of how much potential we had to shape the narrative from the outset. Instead, AI is simply mirroring the biases present in our current society.
Ironically, my initial concern with AI was that it would become overly intelligent or excessively specific. However, I now recognize that specificity is precisely what I desire. I seek personalized experiences and interactions, and I want AI to be tailored to my unique needs and characteristics.
I attempted to generate an image using AI with explicit prompts to include my disability. Yet, AI struggled to accurately depict me. Consequently, I inquired why it was so challenging for AI to create an image that simply reflected who I am.
The response:
We have the opportunity to reshape this narrative. As AI continues to evolve, we must prioritize diversity and disability in its development. This is a crucial moment to alter ingrained social constructs. It is an extraordinary and exciting opportunity, but only if we act now.
Imagine AI-generated images that depict people using wheelchairs, prostheses, or other visible indicators of disability without requiring specific prompts. Envision AI providing this option to everyone by default.
Consider the impact of training AI to challenge societal perceptions of disability, not by request, but by design. Bias arises from the assumption that everyone seeks to “improve” their image, and disability is often overlooked because it is viewed as “less than.”
Disability is not a problem to be solved but a perspective to be valued. By incorporating lived experiences into the design process, AI can move beyond reflecting societal norms and actively reshape them, creating a world where everyone is seen, heard, and authentically represented.
AI should not merely reflect our society; it should contribute to building a more inclusive one. The result will be not only better technology but also a more equitable society where innovation benefits all.
“`