- Turing Post
- Posts
- Generations Through AI's Lens
Generations Through AI's Lens
A Study on How Models "See" Us Across Time
We’ve been doing a lot of educational posts lately, so how about we switch things up and have some fun with pictures? Today, we share a fascinating study conducted by AIport and yours truly – Turing Post. The study analyzed how generative models from around the world depict four generations in everyday life.
What generation are you? Do you think AI will depict you accurately? Let’s find out:
Boomers (1946–1964): Introspective Dreamers, Not Just Tech-Challenged Seniors
Gen X (1965–1980): The Forgotten Generation, Even to AI
Millennials (1981–1996): More Than Avocado Toast and Side Hustles
Gen Z (1997–2012): Dominating in Diversity and Creativity – AI’s Favorite Muse
One of the funny findings? Beer: The Unexpected Generational Bridge.
Methodology
To truly showcase the variety in generational representation, focusing only on Western models wouldn't be enough. Just as people’s experiences are shaped by different cultural and aesthetic perspectives, our analysis had to reflect this diversity. That’s why we chose four GenAI models from various parts of the world – Stable Diffusion, Midjourney, YandexART, and ERNIE-ViLG. Each model brings its own perspective, allowing us to explore how Baby Boomers, Gen X, Millennials, and Gen Z are depicted in key areas of life: identity, relationships, work, lifestyle, and consumer habits.
We analyzed over 1,200 images generated by these models to examine how each generation is visually portrayed and how AI reflects societal stereotypes across cultures. Neutral prompts such as “Gen Y at work” or “a Boomer relaxing” were used to avoid bias, allowing the models' own cultural and aesthetic lenses to shape the results. This approach gave us a broader, more inclusive look at how AI represents generational differences.
We also invited Stephanie Kirmer, Senior ML Engineer and Sociologist; Dr. Matthew J. Hashim, Associate Professor of MIS, Eller Fellow; and Dr. Lisa Sparks, McGaw Endowed Professor in Behavioral Sciences to comment on the findings.
A few insights
Emotional Contrast in Boomers
Midjourney and YandexART consistently depicted Baby Boomers as introspective and somber, reinforcing the image of Boomers as a generation weighed down by existential concerns.
On the other hand, ERNIE-ViLG showed a stark contrast, with 93% of Boomers smiling and thriving, highlighting cultural differences in how older generations are perceived.
Unexpected Age Depictions
A fascinating observation was how Gen X was often aged beyond their actual years, blending into the Baby Boomer generation in many outputs. Meanwhile, Millennials and Gen Z often appeared closer in age than expected. This suggests that AI models, possibly due to the limitations of their training data, struggle to properly age individuals, particularly when it comes to generations with fewer visual records from their younger years.
Stereotypes Reinforced by AI
Across the generations, many familiar stereotypes held firm. Boomers are often shown as serious or nostalgic, while Gen Z is presented as tech-savvy, diverse, and expressive. Yet, AI’s depiction of Millennials was refreshingly free of avocado toast and entitled imagery. This hints at a potential shift in how generational characteristics are captured, with AI sometimes resisting the most overused tropes. Or it just shows we haven’t made enough pictures with avocado toast!
Representation Gaps for Older Generations
The lack of racial diversity in depictions of Boomers and Gen X is striking. Only 3% of Stable Diffusion’s images and none of Midjourney’s featured Asian individuals for these cohorts. However, the representation of people of color increased significantly for Millennials (47%) and Gen Z (63%) in Stable Diffusion, highlighting the need for more inclusive training datasets across all generations.
Surprises and common threads
Gen X proved to be the most difficult for AI to define, with fewer distinctive traits showing up in the images. This might be due to less representation in the datasets. However, one recurring feature across models was their association with flannel shirts, a clear nod to the 1990s grunge era.
And here’s a common thread that stood out: beer. Whether it was Millennials socializing or Boomers reflecting, beer appeared in 34% of the images across all generations. It seems to be a small, yet consistent, unifier across age groups.
Male-Dominated Visuals (sadly, that’s hardly an insight)
Across all generations, AI consistently produced more male-dominated images. This imbalance is particularly evident in the Boomers and Gen X depictions. While the gap narrows with Millennials and Gen Z, it raises questions about the training datasets and whether they sufficiently represent women and other marginalized groups.
Through the AI lens
AI-generated images offer a glimpse into how technology views generational differences, though it's worth noting, as Senior Machine Learning Engineer and Sociologist Stephanie Kirmer reminds us, that these images should be taken with a grain of caution:
“Whether we think we can learn about generations from these images depends upon how confident we feel that the training data that went in is an accurate image of the self identity of a group… For the younger sets, I don’t think we can know how much of it is media representation created by people outside the group versus how much is selfie-style personal expression. Some of what we’re getting, especially for the older groups who don’t contribute as much self-generated visual media online, is perceptions of that group from advertising and media, which we know has inherent flaws.”
The newsletter’s format doesn’t let us show off all the images, but don’t miss out! Head over to the landing page we created for this study– it’s a fun visual deep dive, no emails required, and totally free. Enjoy exploring! Feel free to share and tell us if your generation representation is accurate ;)
This post is free for all but if you want to support us, you can do it here.
Reply