I Used AI to Vote. It Made Me Less Biased, Not More.
The first question my clients ask is "Isn't AI biased?" This election, it was the thing that checked my bias.
When I talk with clients about AI, one of the first questions is almost always the same: "Isn't AI biased?"
It's a fair question. But this election I had the opposite experience. AI didn't push its bias on me. It helped me notice my own.
How I set it up
There was nothing fancy about it. I opened ChatGPT and did two things.
First, I told it who I am. I shared the issues I care about and what matters most to me in a candidate or proposition. Then I gave it my exact location, so it would focus on the measures on my ballot.
Second, I asked for a report on each issue. I just typed the measure number. It came back with what would change, the strongest arguments on each side, and who supports and opposes it.
That was it. Each report took about thirty seconds to read. If you want to try it, here's a starting point:
I vote in county, state. Here's what I care about: your values. When I give you a measure on my ballot, explain what it would change, give me the strongest arguments for and against, and tell me what I might be missing.
Should Boulder County have 5 commissioners instead of 3?
Let me start with a non-partisan example. Boulder County is voting on whether to grow its board of county commissioners from 3 to 5. It's not a topic I think deeply about, and I didn't have a strong opinion. But I knew someone cared enough to put it on the ballot.
Here's what ChatGPT told me:
Boulder county 200
Boulder County Question 200: Expand county commissioners from 3 to 5
This would increase Boulder County's elected commissioners from three to five, beginning in January 2029.
| Current | If 200 passes | |
|---|---|---|
| Commissioners | 3 | 5 |
| Residents per commissioner | ~110,000 | ~66,000 |
| Annual additional cost | $0 | Under $400,000 estimated |
- More elected representatives for 330,000-plus residents.
- More perspectives on housing, transportation, land use, and spending.
- Less concentration of decision-making among three people.
- Two more commissioners add salaries, benefits, and admin costs.
- More commissioners could make decisions slower.
- Boulder and Longmont might gain influence over smaller communities.
The interesting part: Question 201. Question 200 determines how many commissioners Boulder County has. Question 201 determines how they're elected, but only if 200 passes.
What do current commissioners think
The three current Boulder County commissioners are split 2–1 against expanding the board.
| Commissioner | Position | Main reason |
|---|---|---|
| Ashley Stolzmann | Supports | More capacity, better constituent access, broader representation |
| Claire Levy | Opposes | Higher costs, slower decisions, less representation for smaller communities |
| Marta Loachamin | Opposes | Adding seats doesn't fix barriers to political representation |
Levy's financial concern is concrete: the county is cutting staff because of budget shortfalls. She questioned whether adding two commissioners would mean sacrificing five or six staff positions that deliver public services.
That follow-up told me something I didn't know. The sitting commissioners were split 2–1 against expanding their own board. One pointed out that the county was already cutting staff. Another argued that adding seats doesn't fix the real barriers that keep new candidates from running.
It also pointed out that a second question on the ballot decided how the new commissioners would be elected, by district or countywide. I would have skipped right past that.
I wasn't going to spend a day researching county governance. But thirty seconds of reading got me from a gut reaction to an informed vote.
Life without parole for child sex traffickers
Proposition 133 would make life in prison without parole the mandatory sentence for child sex trafficking in Colorado.
This one was heavy. My first reaction was simple: these aren't people I'm inclined to give a break to.
133
Proposition 133: Life without parole for child sex trafficking
This would require a mandatory life sentence without parole for anyone convicted of trafficking a minor for sexual exploitation.
| Current law | Proposition 133 | |
|---|---|---|
| Prison sentence | Generally 8–24 years, potentially longer | Mandatory life |
| Parole | Possible | Not permitted |
| Judicial discretion | Judges can consider circumstances | No discretion to impose a shorter sentence |
Arguments for Yes. Supporters argue that child sex trafficking is an exceptionally serious crime and that permanent incarceration prevents convicted offenders from trafficking additional children.
Arguments for No. The most significant concern comes from organizations that work directly with trafficking survivors.
Trafficking victims are sometimes coerced into recruiting or transporting other victims. If convicted under this proposal, they could face mandatory life imprisonment.
Colorado has some legal defenses for trafficking victims, but opponents argue those protections may not adequately cover every coerced person.
That just-in-time information changed my thinking. I went from "How do we punish these people?" to "What does it look like to treat someone who was a victim before they became an offender?"
It helped me build empathy for a group I knew almost nothing about, which is something I rarely get from a yard sign or a mailer.
Why I think this is the future
There's a lot of worry, much of it fair, that AI shortcuts our thinking.
But it can also shortcut something else: the work of finding the right information.
For most down-ballot measures, the honest choice was never "deep research vs. AI." It was "AI vs. guessing." I was never going to spend a day fully researching the county commissioner question. Without help, I would have voted on vibes or left it blank.
"The honest choice was never deep research vs. AI. It was AI vs. guessing."
Used this way, AI doesn't think for me. It clears away the hours of digging so I can spend my energy on the actual thinking: weighing tradeoffs, testing my assumptions, and noticing where my first reaction was wrong.
The possibilities, and the dangers
For now, I love this. I'm more informed, and I think more deeply about the issues than I did before. But there are risks that worry me too.
It's one click away from "just tell me how to vote." The same tool that lays out both sides will also happily fill in your ballot. That's the scary part: thinking abandoned completely, handed off to a machine. What worries me even more is that the AI would probably do a better job than my uninformed guessing.
It depends on the AI actually giving both sides. In my experience, it did a phenomenal job when I asked for both sides. Each report gave real weight to arguments I disagreed with. But that's a choice the model makes, and corporate interests could change it.
The deepest fear is that most people won't ask for both sides. They'll ask for confirmation. Then again, that's already how a lot of voting works today. We read the mailer from the side we already agree with, ask the friend who thinks like us, and follow the party line. AI doesn't create that habit. It just makes it easier, or makes it easier to break.
AI can be an echo chamber, or it can be the most patient, balanced civics teacher you've ever had. The difference comes down to how we ask.
Nonetheless, I'll be teaching my child how to use AI to think critically about his votes. Though in 17 years, who knows what AI will be like.
If this landed, forward it to someone who's still staring at their ballot.
I build small AI tools for specific people. Each build comes with a video and the thinking behind it.