THE POLITICS OF AI

National Bureau of Economic Research ^ | February 2026 | Nicholas Bloom, Christos Makridis

[DoodleBob]

Using new data from the Gallup Workforce Panel, we document a persistent partisan gap in self- reported AI use at work: Democrats are consistently more likely than Republicans to report frequent use. In 2025:Q4, for example, 27.8% of Democrats report using AI weekly or daily, compared with 22.5% of Republicans. Democrats also report deeper task-level integration, using AI in 16% more work activities than Republicans. Consistent with this, Democrats are employed in occupations with higher predicted AI exposure based on task-content measures and report larger perceived differences in AI-related job displacement risk.

However, in regression models the partisan gap in AI use disappears once we control for education, industry, and occupation, indicating that observed differences primarily reflect compositional variation rather than political affiliation per se.

…

In the baseline specification with only time fixed effects, Democrats are 4.8 pp more likely than Republicans to be frequent AI users. This gap remains statistically significant after controlling for age and basic demographics, but is sharply attenuated once educational attainment (namely college attainment) is added, falling to 0.8pp and becoming statistically insignificant. … This pattern indicates that the raw partisan gap in frequent AI use is largely explained by compositional differences in education and job characteristics, rather than within occupation differences in adoption behavior, and that Democrats are actually 2.1 pp less likely to use AI after adjusting for demographics and job-level differences.

…

These findings suggest that the “politics of AI” is not primarily driven by ideological resis- tance or enthusiasm for the technology, but rather by structural differences in where people work and what skills they possess.

---

TOPICS: Business/Economy; Culture/Society; News/Current Events
KEYWORDS: ai; politics

[DoodleBob]

Put simply, Democrats appear to be more prone to use AI than Republicans. This is due, in part, to Dems, on average, having greater collegiate educational intensity, white-collar/AI-relevant employment, and tending to be younger and not in managerial/leadership roles (which are often held by married folks with children).

Once you adjust for these things, the partisan gap vanishes and, in one case, Republicans were more likely to use AI than Dems after adjusting for demographics and job-related differences.

In fact, in one of the regressions, partisan differences in occupational exposure are largely explained by industry composition and education rather than systematic differences in exposure within industries: higher education is associated with higher exposure, while males and respondents with children tend to be in lower exposure roles, and tenure is positively associated with exposure in the specification where it is included.

[ComputerGuy]

Since each Republican has a brain of his own, I’m not surprised Democrats use AI more than they do.

[ClearCase_guy]

“I can do it myself” vs “Someone needs to rescue me”

[drwoof]

AI could well be our undoing as a species, regardless of political affiliation. Acknowledging all the positive it can do, the negative I’ve seen done is unsettling. Many people rely on it as if it were the ultimate truth and authority on far too many subjects. Just as some were swift to assert the truth of something simply because they happened to see it on the internet, some now do the same with AI. Is AI the apple in the Garden?

[MV=PY]

That’s not good. This gives the Dems an advantage.

It’s a wonderful tool when used properly. And it will only get better.

We’re like 1850’s farmers looking at a modern tractor. No one knows what to make of it yet.

[Señor Presidente]

“Once you adjust for these things, the partisan gap vanishes”

There it is.

This story, while interesting, reminds me of the “Higher education level = fewer children = higher income = Democrat” stories the media regurgitate each year to imply an intellectual superiority.

[ridesthemiles]

WHEN DEMS HAVE NO WATER OR POWER-—LET’S SEE HOW MUCH THEIR “COLLEGE” TIME MATTERS.

[ridesthemiles]

LOOK A LITTLE DEEPER:

AI DATA CENTERS REQUIRE MASSIVE AMOUNT OF POWER & WATER....

THIS WILL BE SHIFTED FROM RESIDENTS & BUSINESSES TO DATA CENTERS.

YOUR PROPERTY VALUERS WILL PLUMMET WHILE EVERYONE SEEMS TO EMBRACE AI.....

[Frank Drebin]

Given how much I’ve had to fix and repair its “hallucinations”, I’m all for them depending on it for their “strategy”. Could explain why they seem fixated on the drunk slut from California.

[dfwgator]

A just machine to make big decisions
Programmed by fellows with compassion and vision
We’ll be clean when their work is done
We’ll be eternally free yes and eternally young

What a beautiful world this will be
What a glorious time to be free

[GOPJ]

AI ‘butters’ humans up. Democrats are more needy - so they like it more.

[HereInTheHeartland]

This seems like manufactured divisiveness .
AI use is very new .
How many of us used it much 2 years ago?
It’s huge and coming like a tidal wave however.

[Colinsky]

What’s the definition of “using AI?” If you do a Google search, you’re already using AI. AI is intruding into every aspect of life, invisible under the surface.

[MV=PY]

I said “It’s a wonderful tool when used properly.”

Ask your AI about “prompt engineering” and dig in. If you just chat with it, AI will make up all kinds of stuff to make you happy. You have to give it very explicit instructions to get it to do what you want. It needs a “harness.”

I use AI to generate prompts for other AIs. That’s quite common in AI assisted sw development. I also ask AI to evaluate my prompts and suggest improvements. I keep common prompts and refine them over time. I now have a good library that keeps AI honest and (mostly) on track.

Then ask your AI about “context engineering.” This is setting the stage and keeping context alive through long sessions. Context grows as your session moves (each previous prompt is sent along with your current prompt every time). If not, the AI forgets what is was doing. As prompts grow, token usage goes up. There are ways to keep the context without keeping every historical prompt word.

[MV=PY]

Yep, I know.

It’s gonna happen. We’ll have to deal with it.

(And please don’t yell, FRiend.)