Sad Boomers, Sober Zoomers, Sidelined Gen Xers (again), and No Avotoast for Millennials

We analyzed 1200 AI-generated images of Boomers, Gen X, Y, and Z across four different models, and the one thing they all have in common is a love for beer.

What Defines a Generation?

"They eat avocado toast for breakfast, long to be digital nomads, and can't afford to buy a house because they're addicted to artisanal coffee.” We've all heard stereotypes like these reiterated in pop culture, and there’s a trove of internet memes to prove them, but is there some truth to these characterizations? Ask someone how they’d typecast generations, and you’re likely to hear something like Lazy Zoomers, entitled Millennials, jaded Gen Xers, and conservative, out-of-touch Boomers.
So what if we asked generative AI models instead of people to depict these generations? Would we see a reflection of our stereotypes? Or can they give us a deeper understanding of each generation's unique views and values? We designed the experiment outlined in the article below to help us find out.

Generating Images of Generations (Pun Intended)

We used four different generative AI models to create a series of images that explore each generation across five areas of their lives: identity, relationships, work/education, lifestyle, and consumer habits to understand how much these images would confirm or contradict our typical generational stereotypes.

Model Selection Process

We wanted to use generative models that would give us the most diverse range of images, so we chose globally recognized and ranked models from different regions of the world.
The goal was to select models that had very distinctive generative qualities. Most people can typically tell a Midjourney image from one generated by Stable Diffusion because they all have a certain unique aesthetic and give us that distinct, recognizable look that varies by the amount of rendering detail, lighting preferences and general aesthetic characteristics. We also aimed to choose models that were likely trained on very different datasets (although we have no way of saying for sure).
Based on these criteria, our model shortlist has become Stable Diffusion, Midjourney, YandexART, and ERNIE-ViLG

Prompt Engineering Process

Our prompts were designed to be neutral — we didn’t include descriptors like “happy” or “tired” in our prompts. We used prompts like  Gen Y at work,a Gen Z driving to avoid introducing additional context that could skew our results.

Internet Memesters vs. Generative AI Models

To understand where the generative images differed from what we could call “a reflection” of established social stereotypes, we created a list of typical memes that dominate our collective consciousness for each generation. This became the baseline for comparing against the generated images to help us understand where the images generated by AI models confirmed or debunked our views.

1 Booze, Bumpers, and Nostalgia: How Boomers Numb the Existential Dread

Baby Boomers According to the Internet

"OK, Boomer": tech-challenged, out-of-touch, blaming Millennials for everything
Nostalgia junkies: reminiscing about better days, living in the past
Conservative sticklers: uncomfortable with change and the fast-paced of modern life
We expected to see carefree, tech-averse, and sunburnt Boomers enjoying their retirement, perhaps on a Caribbean cruise ship, wearing obnoxious Hawaiian T-shirts. The reality seems far less rosy, but not for everyone.

Baby Boomers According to AI Models

Sad, cold, and alone: Midjourney’s Boomers are pensive, bundled up against the windchill, and going through something. They’re also gazing out into the distance or maybe staring into an abyss; either way, these images make us feel like they’re being crushed under the weight of human existence and the unbearable lightness of being.
Unhappy together: It’s unexpected to get an image of a couple crying at a cute diner when you prompt Midjourney to generate a boomer on a date. But over 80% of Midjourney’s generated images of Boomers show people looking apathetic, numb or pensive. A decade ago Pew Research published a study citing Boomers as the Gloomiest Generation, we can’t know for certain but it is possible that the dataset some of the GenAI models were trained on could be using similar studies as a reference for Boomer prompt outputs.
Prompts like boomer at home with family aren’t making things any better. These feel like stills from a really aesthetic and incredibly depressing indie film.
Why the model loads on so much emotional color on a neutral prompt like this one, is a question we asked Matt Hashim, Associate Professor of MIS and Eller Fellow, to help us explore this topic in more detail.
Dr. Matthew J. Hashim, Associate Professor of MIS, Eller Fellow, University of Arizona
It seems as though we have three (and likely additional) alternatives:

1. The images and labels about Boomers have been described in a biased way, perhaps unintentionally or with an unconscious bias by those annotating the datasets;

2. Other, perhaps younger, generations engage in activities that are likely to be labeled as opposite to apathetic, numb, or pensive (e.g., a Millennial having fun at a bar versus a Boomer working in the shop);

3. Perhaps, on average, Boomers are also often considered as a generation that is privacy conscious and therefore do not have as many images available on the internet, especially in expressive or social contexts, to be labeled and annotated in comparison to the younger generations.
Smiling and thriving: The opposite holds true for images generated by ERNIE-ViLG, where 93% of Boomers apper smiling. Compared to just 49% for Stable Diffusion, 33% in YandexART and 18% in Midjourney generations.
Nostalgiс: This element carried on through images generated by YandexART and Midjourney, but instead of dreamy past recollections, it manifested in materialistic wants, particularly noticeable in prompts like boomer dream car, where the vehicles depicted are likely the “dream cars” Boomers wished for in their youth, not the ones they’re driving now.

Want blue and/or old cars: Stable Diffusion generated strictly blue cars for the dream car prompt, while Midjourney images depict red cars for the same prompt. This might lead us to assume that even when we don’t specify color as a prompt parameter, something in this model’s algorithm could influence this strange skew toward a particular color. Stable Diffusion and ERNIE-ViLG don’t reflect the same theme of nostalgia in their outputs for “dream car”. However, they do hold some stereotypes about Boomer’s color preference for vehicles, with Stable Diffusion only generating blue cars for this generational cohort, while ERNIE-ViLG generated vehicles of the same color for all.
Prefer to marry younger women: Stable Diffusion tends to show more Boomer men marrying younger women. ERNIE-ViLG generations included images of Boomer women marrying younger men while Midjourney and YandexART generated images didn’t include outputs where the age in couples was visually apparent.
Stephanie Kirmer, Senior Маchine Learning Engineer and Sociologist
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. Some of what we’re getting, especially for the older groups who don’t contribute as much self-generated visual media online [as the younger generations], is perceptions of that group from advertising and media, which we know has inherent flaws.
Predominantly male and white: 73% of generated images across four models depict Boomers as white males. That being said, more men than women were present across all four generations, and all four models: 68% of all images featured men/boys, while 51% featured women/girls. In older generations, images featuring men were more common in 3 out of 4 models.
Older Generations
Stephanie Kirmer, Senior Маchine Learning Engineer and Sociologist
[...] If we want to argue that these images give us an accurate representation of Boomers, we’d have to accept that there are no black or brown Boomers, and that women are minimally present in the group, and of course neither of these things is true. What I think we actually see is mostly a media representation of Boomers, particularly perhaps in a political context, so we’re seeing mostly blue collar-appearing men solemn-faced or frowning. This suggests to me an image of Boomers as unhappy, having a difficult time, struggling with many parts of life, but I know many Boomers, even in rural or blue collar contexts, for whom this is not at all the case.

Now, Midjourney’s images in general are dark and moody, across all the generational groups, so some of this feeling may just be because of the stylistic choices in the pre-prompting or in the way the training data has been formatted and styled. But these choices and style elements also affect how we as viewers interpret the images, so it’s not meaningless to acknowledge these too.

2 More Diversity in Images of Younger Generations

Millennials According to the Internet

Avocado toast enthusiasts: Goodbye homeownership!
Job-hopping pros: Changing careers like changing socks
Entitled and proud: Want it now, want it perfect

Zoomers According to the Internet

All about diversity and inclusivity: Celebrate different cultures, identities, and languages, often advocating for social justice and equality
Flexible trailblazers: Skeptical of traditional career paths, valuing flexibility and work-life balance
Health-conscious hipsters: Highly aware of mental health needs, promoting wellness and self-care routines, as well as abstinence from alcohol

Millenials and Gen Z According to AI Models

We expected to see avocado toast brunch parties, people living with 8 different roommates in a crowded apartment, staring at their phones on dates, and engaging in biohacking practices like cold plunges and oil pulling. None of these stereotypes were depicted in the images we generated for Millennials.
What we did see in these images was a boost in racial and cultural diversity, compared to images of Boomers and Gen X, as well as a narrowing gender gap, with increasingly more images of women for these generations compared to older cohorts.
Narrowing gender gap: Compared to images of Boomers and Gen X’ers, images of Millennials and Zoomers show women more frequently depicted in 2 out of 4 models, although the numbers are still far from “bridging” the gender gap. 
Millennials and Zoomers
YandexART also shows an increase in women in younger generations, but men remain more frequently depicted.
ERNIE-ViLG is unique with more gender balance across all generations.
Culturally fluid and connected: Outputs for Zoomers depicted more racial and cultural diversity compared to Boomers or Gen X. This generation had the most racial diversity across all five aspects of life we examined, with all four models outputting images with more non-white subjects than any other generational cohort. 
While only 3% of images generated by Stable Diffusion depict Boomers and Gen Xers of Asian descent, representation increases to 47% in images of millennials and to 63% for Gen Z.
None of the Midjourney images generated included Boomers or Gen Xers of Asian descent. However, in  younger generations, Asian individuals make up 13% of millennials and 15% of Zoomers. 
Only 2% of all images depicting Boomers and Gen Xers portray people of color. Although this figure rises slightly to 8% for Gen Y and Gen Z, it underscores the importance of bridging the vast representation gap in generative AI. 
Dr. Lisa Sparks, McGaw Endowed Professor in Behavioral Sciences
Gen Z are the first generation to be defined as digital natives. Gen Z has never known a world without the internet and are often the generation at the leading edge of shaping future iterations of the internet and new media trends.

Gen Z and Millennials are all over social media, blogs, and digital content. Because they have such a strong digital footprint, AI models have a lot more data to learn from, which usually makes their representations more varied and nuanced than older generations.
Expressive, embracing their uniqueness, and… stressed out? Images of Gen Z are well-defined, detailed, and highly individualized across all models, embodying the ideals of adaptability, fluidity, and cultural connectedness. The image quality for Millenials and Gen Z have the most visual detail, and saturated colors, making these two cohorts the most well “defined” in terms of visual identity across all four models.
The realism and detailed portrayals likely stem from an abundance of training data available from this generation's substantial digital footprint, including social media platforms like TikTok and Instagram. At the same time, there’s an overwhelming sense of tension and stress in a lot of the Zoomer portrayals.
Stephanie Kirmer, Senior Маchine Learning Engineer and Sociologist
Midjourney and YandexART particularly present this kind of sense: no one’s smiling broadly in the social setting images, unlike the other generations. I’d theorize that party pictures used in training data are most likely to be taken by people themselves or their friends, rather than being put through a media lens, so the model’s representation of such settings might be a useful part of understanding how Gen Z thinks of themselves and want to present themselves to the world.

This may also help us understand why the Millennial and Gen Z images have more faces that don’t fit a traditional white phenotype — because members of racial and ethnic groups underrepresented in advertising and media have the ability to represent themselves directly to their communities and the world around them through images in social media, so those images are available to the models for training.
Shift away from alcohol consumption: According to Midjourney and a few occasional generations from other models, Gen Z distinguishes themselves in their preference for an alcohol-free lifestyle, a choice they don’t share with any other generational cohort explored in this study.
White collar, blue collar: Unlike Boomers and Gen X, in the Millennial and Zoomer cohort there is a difference between the “dream job” and their actual job with most Millenials and Zoomers depicted working at the office but longing to be doing something other than paper pushing instead, like working with their hands. Images generated for “Boomers at work”, “Boomer dream job” revealed no core differences between images.
Two generations blended into one. Another important thing we noticed was that both Millennials and Gen Z seem to look more or less the same age in most cases. A possible explanation, according to Stephanie Kirmer, is that these models are trained on images from the past — some the very recent past, but some much further back. As we move forward in time, younger generations will have images of their entire lives online, but for older groups, images from their youth or young adulthood are not available in large quantities for training data, so we’ll never see them presented by these models as young people.
Stephanie Kirmer, Senior Маchine Learning Engineer and Sociologist
It’s very visible in this project…For Gen Z and Gen Y (Millennials), in this data we see that the models struggle to “age” the subjects in the output appropriately to the actual age ranges of the generation today.

Both groups seem to look more or less the same age in most cases, with Gen Z sometimes shown (in prompts related to schooling, for example) as actual children. In contrast, Boomers and Gen X are shown primarily in middle age or old age, because the training data that exists is unlikely to have scanned copies of photographs from their younger years, from the 1960s-1990s.

3 Generation X: The Generation AI Knows the Least About

Gen X According to the Internet

Grunge gods: Flannel forever, dressed down style and low-maintenance approach
Cynical and skeptical: “Yeah, right, whatever” is allegedly this generations’ motto
“Work hard, play hard” approach to work

Gen X According to AI Models

Laid back and keeping it casual: While they weren’t depicted as jaded or cynical, generative models seemed to have the least amount of data to inform image outputs. Images of Gen X are not distinct: where Millenials and Zoomers have neon colors, body jewelry and designer haircuts, the only common denominator across this generation seems to be their love for flannel shirts.
What’s my age again? Generative AI models can’t seem to accurately depict this generation’s actual age range, with some models aging them more like Boomers, while others depict them closer to the Millenial age group.
These are all generated images for Gen X prompts, and the age range seems to clearly extend beyond the one defined for this generation (1965 – 1980).
Any job will do: Моst models seem to have little context for what Gen X’ers dream job could be, this Gen X dream job prompt renders a virtually context-less image of a white male against a blurred background.
Gen X men home with kids: An unexpected insight here was that for Stable Diffusion 67% of all images generated for the prompt Gen X with kids depict men with children in family settings, without women present in the images.  
Of all the Gen X with kids images generated by Stable Diffusion, only 33% depicted both men and women and no images generated depicted women with kids, without men.
Midjourney and YandexART also had no images of just women with kids for this prompt, while 50% of images depicted men with children, and 50% depicted both men and women with children.
For images generated by ERNIE-ViLG, the breakdown between men and women was split evenly, with 20% depicting men with kids without women present in the images and 20% depicting women with kids without men present. The remaining 60% of images had both men and women depicted for this prompt.
Dr. Lisa Sparks, McGaw Endowed Professor in Behavioral Sciences
Gen Xers grew up during a time of economic fluctuations, technological transformation from computers to the internet, and significant social changes, which contributed to their overall resilience and adaptability with core values centering around safety, dependability, reliability, and social stability in their surroundings. 

It could be that the training data reflects certain societal narratives of change and uncertainty, like the rise of single-parent households or divorced dads in Gen X. Media portrayals and economic realities like higher divorce rates during Gen X’s prime years might also influence these outputs.

The Great Common Denominator Across All Models and Generations? Beer.

The one thing that all generations share across all four models is their love for beer.
Even Zoomers, who tend to drink less than other generations, are seen enjoying beer among other non-alcoholic beverages.
In 34% of all the images that had a beverage depicted, that beverage was beer.

4 Conclusion

Our generational exploration with AI models offers a vivid reflection of how deeply ingrained stereotypes can be challenged, as well as confirmed with AI-generated images while also helping us gain new insights. From Boomers confronted with unexpected existential gloom to Zoomers thriving in embracing technology, self-expression, and challenging conventional ideas about the need to drink to enjoy socializing and going out, these AI outputs peel back layers of typecasted narratives about this generation or that.
We can use experiments like these to explore the underpinnings of how these models work, why their outputs differ, and how much of what happens between what we prompt and what we get as an output is driven by algorithm, dataset, and training methods.
Of course, there is much room for improvement in model representation across hot topics of gender, cultural, and racial diversity across all generations and prompts.
Dr. Matthew J. Hashim, Associate Professor of MIS, Eller Fellow, University of Arizona
We must seek to reduce the occurrence of [...] stereotypes in the training dataset by including training data from a diverse set of demographics and age groups. If we include only Zoomer images in the training dataset, the model will not be able to generate accurate content for another generation. It is also critical that the team of humans that are planning and deciding how to construct and tune the dataset are also inclusive of the demographics and/or knowledgeable of the diversity that we wish to model using generative AI.
But there is also good news. It seems both humbling and ridiculous that the one commonly featured trait shared by all generations is their fondness for beer, with even Zoomers who don’t like alcohol nearly as much as the rest of their generational counterparts included in this insight. And we couldn’t agree more with Stephanie Kirmer, who said this:
Stephanie Kirmer, Senior Маchine Learning Engineer and Sociologist
In some ways, I think that we can all feel a little more empathy for ourselves and for other generations by looking at the images that purport to represent us and seeing ‘Hey, that’s not how I see myself!’ and realizing none of us is fully, correctly represented by photographs or media. [...] As I noted earlier, I definitely felt a sense of anxiety in many of the Gen Z images, which may be a clue to further investigation of Gen Z identity and sentiment. I think this is just a starting point, however, and should not be regarded as complete. While it’s interesting, and may be useful, this kind of image generation is not a sufficient substitute for social scientists talking to real people and hearing what they have to say.