"A Beautiful Woman"
AI's image generated from the request for an image of "a beautiful woman"
AI's image generated from the request for an image of "a beautiful woman"
If you ask Artificial Intelligence for “a beautiful woman” ten times, you will get ten white women, all thin with alluring eyes. She will most often appear with straight hair and full breasts. Ask AI for a group photo of beautiful women, you will get the one above, an image containing only one aesthetic.
If you want AI to produce a woman of color, you have to ask specifically for her in your prompt. You will generally not see pigmentation variety otherwise.
In the enchanting world of Artificial Intelligence, we all seem to lose ourselves in the terror and promise of its outcomes and potentials. However, reflection on the inherent bias in artificial intelligence is paramount to rectifying harmful outcomes.
1.
Perpetuation: AI's Role in Bias and Stereotypes
"A Beautiful Woman"
Stephen Cave and Kanta Dihal posit in "The Whiteness of AI" that images like the ones generated by Artificial Intelligence "simply reflect the predominantly White milieus from which these [artifacts] arise." As images are crafted from the while male gaze, concepts like "a beautiful woman" are presented in ways that perpetuate limited views of beauty. Furthermore, Cave and Dihal astutely note that "[t]he power of Whiteness’s signs and symbols lies to a large extent in their going unnoticed and unquestioned." Scholar Leon Furze acknowledges that "Beyond the narrative of cheating and dramatic claims of the end of industrial-era education, AI offers real opportunities to personalize learning and help with differentiation, diversity, and inclusion. But there is a shadow side to AI which cannot be ignored."
Therefore, we must begin questioning the output generated by "predominantly White milieus" and cease to ignore what the products perpetuate.
“All data are biased, and you can approach tools in an ethical and responsible way.”
2.
Socially Responsible: Developing Conversations and Questions around AI Imagery
In higher education, as well as within learning at all levels, the ability to access AI is quite simple. We are generally only a few clicks away from an essay response, an html code, or a new image to embed.
However, as students explore what AI can be, it is important for educators to help mitigate the potential harm AI can bring by teaching students to question the inherent bias in the AI algorithms, harnessing the power and product of AI as a tool for the greater good. As Belkacem Boughida notes, "All data are biased, and you can approach tools in an ethical and responsible way." How then do we cultivate the ethical and responsible application?
We can begin by asking ourselves "will artificial intelligence..." "is artificial intelligence..." and "should artificial intelligence..." Through that questioning, via deliberate conversations surrounding AI, the process begins.
Will AI ever provide unbiased results?
Is AI capable of presenting users with valid outcomes?
Should AI produced products be trusted?
"A Beautiful Woman"
"A Beautiful Woman"
"A Beautiful Woman"
"A Beautiful Woman"
"A Beautiful Woman"
Knowing that "A Beautiful Woman" is only one output for AI, as educators, we must process those same questions. "Will artificial intelligences provide me with unbiased results?" "Is artificial intelligence's result socially acquired?" "Should artificial intelligence be trusted?"
We must encourage our students to ask the same questions and begin demanding the variety of results that are necessary to be truly reflective of our world.
"A Beautiful Woman"
3.
Only by Request: Helping Break Bias by Demanding More
According to Susan D’Agostino "AI outputs are often not representative of the depth and breadth of many students’ multicultural and multilingual experiences" and when our students don't seem themselves or a reflection of the greater community in the output, problems arise. We can not let those biases prevail, so our lessons and discussions in the classroom can begin demanding more from AI. Kansas State University notes that"By better understanding the many facets of bias and considering how it influences our actions, though, we can improve our ability to lessen its effects."
All images of "beautiful women" on this site are products of generative artificial intelligence. The following websites were consulted for the production of said images: https://hotpot.ai/art-generator and https://deepai.org/
The Google prediction images are screen shots from the author's personal search, and any other content on this site is linked to the website and source of origin.
Built with Shorthand