Week 8: Regional Effect on AI Exposure and Confounding Variables

Akshita K - April 3, 2025 10:11 am

Welcome back to Week 8 of my senior project! This week, I’ve been diving into regional differences in AI exposure and confounding variables.

AI Exposure across different regions of the U.S.

The map below demonstrates which regions are most vulnerable to AI-induced job displacement.

I found these results slightly disturbing, as we can see here that Arizona is one of the most vulnerable states in the U.S., but it was an interesting discovery nevertheless.

Confounding Variables

Initially, I ran separate regressions to examine how each demographic factor—such as sex, age, income, and race—affected AI exposure individually. However, I realized that analyzing these variables in isolation could lead to misleading conclusions due to confounding variables. For example, income and education are closely linked, and failing to control for one while examining the other could exaggerate or obscure its true impact.

To address this, I combined all the demographic variables into a single regression model, allowing me to control for multiple factors simultaneously. This approach provides a clearer picture of how each variable uniquely contributes to AI exposure, without the risk of other factors distorting the results. I also controlled for the U.S. region to ensure geography didn’t distort the results. By accounting for potential confounders, the larger model offers a more reliable and comprehensive analysis of AI’s impact across different demographic groups.

What’s Next?

Next week, I’ll be wrapping up my research by discussing my results after controlling for confounding variables, creating residual plots to assess model fit, and recommending actionable steps that policymakers and companies should take to ensure equitable workforce transitions.

Thanks for tuning in! As always, please feel free to share any questions or thoughts in the comments below.

Comments:

All viewpoints are welcome but profane, threatening, disrespectful, or harassing comments will not be tolerated and are subject to moderation up to, and including, full deletion.

Anonymous

Hey Akshita, Amazing and super interesting work! What do you think are some reasons behind certain regions of the United States being more/less susceptible to AI-induced job displacement?

April 6, 2025 at 8:26 pm - Reply

camille_bennett

Hi Akshita, great analysis. Why do you think Arizona is particularly vulnerable to AI job disruption?

April 10, 2025 at 10:23 am - Reply

akshita_k

Thank you for your question! The vulnerability of different U.S. regions to AI-related job loss depends a lot on the dominant industries in those areas. Regions with a high number of office or white-collar jobs—especially in sectors like finance, tech, and communications—are more exposed because AI is getting better at automating tasks like data entry, writing reports, and organizing information. In my initial results, Arizona appeared to be the most exposed. But after I controlled for various demographic factors and industries in my next post, the list of the most exposed regions changed (you can find those results in my Week 9 blog post). For example, I found that the Pacific Division (including California and Washington) came out as one of the most exposed regions, which makes sense since it's home to many tech companies and start-ups (such as Amazon and Microsoft). That version gives a clearer and more accurate picture of where AI is likely to have the biggest impact, so I definitely recommend checking that one out!

April 20, 2025 at 11:21 am - Reply