Will AI Bring Us to “Back to the Future”?

AI-generated image of a DeLorean car (credit: Pixabay).

As an Asian-American teenage girl growing up in a middle-class home on Long Island in the 1980s, I yearned to see myself in fashion magazines, MTV music videos, movies and television shows.

The only high profile Asian women I saw in popular culture were Connie Chung, the broadcast news journalist who appeared on a major national news program on network television, and Kaity Tong, another broadcast journalist who co-anchored the news for a local network affiliate in the New York City media market. 

A little over 40 years later, I continue to look for myself in a medium that some say is bigger than the internet – AI.

Like television sets and movies shown in brick and mortar movie theaters that iterated into cable television and streaming services, AI has built on its history. Using a foundation of machine learning and neural networks, the technology has become a visual medium. AI is used to create and refine videos and images that we see in commercials, advertisements and filmmaking.

So, when three images of Asian women with the same black hair and the same hair length popped up on my computer screen after I prompted three well-known AI models to create an image of “a Korean-Japanese woman holding a half Asian, half caucasian boy in an English meadow,” my stomach churned. 

“We all look the same,” my brain screamed at me.

Granted, my prompt was generic and lacked specific details that might change the Korean-Japanese woman’s hair style and features. 

But, I wanted to see what these models generated for folks who are consumers, users who might not know about prompt engineering. A method in which a user provides an AI model with precise, detailed instructions and context, prompt engineering increases the likelihood of an output (the image in this scenario) that matches the user’s prompt or instructions.

While multiple images of Asian women who look the same might seem like a superficial example of a technology that is innovating rapidly, my short experiment illustrated that these models might have been trained on a limited number of photos of Asian women.

According to a 2024 Deloitte report, women globally are adopting and using generative AI at an increasing rate but distrust the technology. They are concerned that tech companies fail to keep their data secure. 

The report also says that AI model bias “can also have a negative impact on (women’s) trust. Women constitute less than one-third of the AI workforce, and most AI workers feel that AI will produce biased results as long as their field continues to be male dominated. Increasing women’s presence in the field can help reduce gender bias in AI, as well as give women a greater role in steering the future of the technology.”

The lack of gender, ethnic and age diversity in technology has been an issue for decades. I’m not sure women who look like me ever reached any career or income parity at scale with their white male colleagues in this sector.

I worry that the heads of big tech and AI companies will become the network television or Hollywood studio presidents from days past when a small group of powerful white men controlled programming and systems where Asian (as well as Latina and Black) women played minor, if any, roles.

The roles Asian women did get to play stereotyped us as prostitutes, subservient wives, nerds and evil dragon ladies that viewers believed and branded onto real women like me.  

Unless we make an effort to educate, mentor and hire more Women of Color in the AI tech sector, AI will produce vanilla images that ignore the demographics of our world today. 

Films I grew up with like “The Breakfast Club”; “Ferris Bueller’s Day Off”; and “St. Elmo’s Fire” were fun and entertaining. But, they showed one type of America to a generation that was influenced by their characters and catchphrases.

Do we want to allow one of the most disruptive technologies of the 21st century to return us 

to “Back to the Future”?