AI Unpacks Generational Stereotypes, Highlighting Key Differences, but Beer Bridges the Gap

September 23, 2024 at 10:00 AM EDT

--News Direct--

What happens when you ask generative AI models to depict Baby Boomers, Gen X, Millennials, and Gen Z? You might expect tech-challenged Boomers and avocado-loving Millennials. But what you get is far more interesting and in some cases, unexpected.

A joint research project conducted by AIport and Turing Post last month analyzed over 1200 AI-generated images across four different models—Stable Diffusion, Midjourney, YandexART, and ERNIE-ViLG. The findings have revealed both familiar stereotypes and surprising quirks about how AI perceives these generations.

Sad Boomers, Happy Zoomers?

Contrary to the carefree Boomer stereotype, AI models like Midjourney depict Baby Boomers as introspective, or even somber, often bundled up and gazing wistfully into the distance—a reflection, perhaps, of their disillusionment with the unfulfilled ideals of the 1960s cultural revolution. However, the ERNIE-ViLG model, likely trained on datasets with a more collectivist cultural slant, shows 93% of Boomers smiling. This fascinating contrast underscores the cultural differences embedded in different AI models.

Meanwhile, Gen Z’s images are vibrant and diverse, reflecting their reputation as a generation that embraces individuality, inclusivity, and self-expression. The AI-generated images depict Zoomers in colorful, highly detailed scenarios, making them the most visually distinct generation in the study.

Generations Through the AI Lens

The research explored how generations are portrayed in five key aspects of life: identity, relationships, work, lifestyle, and consumer habits. The AI models revealed consistent patterns and stereotypes; for example, males predominated in the depictions of Boomers and Gen Xers across all four models, while many images of Millennials and Zoomers showed greater diversity and more female representation.

Surprisingly, Gen X appears to be the least well-understood generation by AI, characterized by fewer defining features compared to all other generations, likely due to the limited amount of training data available. The one stereotypical Gen X leitmotif that AI consistently identifies is their fondness for flannel shirts—an unmistakable symbol of the 1990s grunge scene worn by Gen Xers who were rebellious teenagers and young adults during that era.

As for one (un)expected commonality… It’s beer. Whether they’re Millennials job-hopping or Boomers reminiscing about the good old days, AI consistently showed beer in 34% of the produced images across all generations.

Prompt Engineering and AI Insights

The prompts for this research were carefully crafted to avoid bias, using neutral phrases like "A Millennial at work" or "A Boomer relaxing." The results offer a telling look at how generative AI models mirror societal stereotypes, often shaped by the data they're trained on. These findings open the door to deeper discussions on how AI reflects, and sometimes distorts, cultural narratives.

"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,” explains Senior Engineer and Sociologist Stephanie Kirmer, who analyzed the findings. “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."

About the Study

The study examined the outputs of four globally recognized generative AI models, each with distinct aesthetic and cultural nuances, providing a comprehensive look at how each generation is portrayed visually. The selected models included Stable Diffusion, Midjourney, YandexART, and ERNIE-ViLG, offering diverse perspectives from different regions across the globe.

Conclusion

Whether confirming or contradicting societal stereotypes, AI-generated images provide a fascinating snapshot of how technology interprets generational differences. The fact that beer remains a unifying theme across all four models shows that, no matter our age, some things—like enjoying a cold one—transcend generational divides.

To learn more about the project, visit https://www.aigenerations.tech/.

Hero images for the media: Google Drive Link

About AIport

AIport is an emerging community of AI/ML enthusiasts and practitioners with a goal of providing a truly global view of the AI/ML/DS landscape.

https://www.blog.aiport.tech/

About Turing Post

Turing Post is everything you need to make smarter decisions about AI.

https://www.turingpost.com/