Can AI Actually Make Strategic Decisions for Your Campaign Right Now?

The quick answer is: “Not yet.” I’ll show you why.

AI is undoubtedly powerful, and you’d be a fool to claim otherwise. It gets better on a fairly regular basis, and as someone who does a lot of coding and long-form writing, I use it all the time. In fact, I used it to write the rough draft of this article. Well, sort of. 

I opened a new chat, uploaded all my previous C&E articles, then jotted down a few paragraphs of thoughts, and finally, told it to ask me 10 follow-up questions. Using AI in this way is a great way to brainstorm, and yet, the first draft ChatGPT wrote was, to put it mildly, AI slop.

The current discourse online about AI writing is around the use of em-dashes, bullet-point lists, and emojis. But I find the most telling, and destructive feature of AI writing to be what folks call “hallucinations”— its propensity to make up random things, often completely contrary to what you’ve told it.  

A major issue, I think, is that AI doesn’t remember details well over long conversations. It trends toward generic insights, even when you prompt it with clear guidelines. It doesn’t connect ideas in a way that reflects your real-world experience because it doesn’t have any real world experience of its own, only what it’s been trained on, and if you’ve read much of the printed materials available on campaigns you’ll understand why that is such a problem. 

Most importantly, it cannot process something as massive as a voter file. But if you take the time to summarize that data yourself, AI can assist in writing the code that will make it into charts then generating questions to ask you about the trends you see within those charts. You still must be the one actually interpreting the results.

Managing a campaign requires you to lean on your experience, be strategic in your

decision-making, and adjust to the surprises that will definitely come up. AI can certainly assist in this process, as a brainstorming tool, but it cannot think for you — and it certainly cannot run your campaign. That’s why you hear folks saying things like “AI isn’t going to take your job, but someone using AI might.” 

If I ask it to generate 20 out-of-the box ideas based on my experience, it might come up with three that are good enough to help me make a good decision. But let me demonstrate. 

I opened a new chat and provided the AI with a common problem: “A large majority of my district is apartment complexes, and it’s difficult for me and my volunteers to reach these folks. Come up with 10 ideas that might help me reach them.” 

Here is what it came up with, and my thoughts on each:

1. Coordinate with Property Management: Why would they let us in? And anyway, I’ve tried sneaking into these buildings before. Access is not the problem, but the fact that apartment owners don’t open their doors if you knock, and even when they do, they’re not happy you’re there.

2. Send Targeted Direct Mail: This is expensive, and doesn’t really allow for the person-on-person connection that I was hoping for from the doors.

3. Host Nearby Community Events: I’ve tried this, and it works okay, but it brings us back to the same problem of how do I make these folks aware of my campaign and the event?

4. Use Digital Advertising Geo-targeted by Apartment Complex: This is the same issue as mail, though maybe a little less expensive.

5. Flyer Drop-offs and Door Hangers: Two problems: Firstly that I cannot access the building without sneaking in, and second that lit-drops are not always very effective.

6. Text Outreach: This is alright, and I think more along the lines of what I was looking for the AI to suggest — an alternative personalized contact method.

7. Establish Apartment Ambassadors: I’ve actually tried this, and it was, well, okay. Of course it required that I find these people in the first place, which required me to do some text or phone-banking.

8. Transportation Assistance for Events or Voting: This one is as bad as the previous two are good. In fact, these large apartment complexes often ARE the polling place.

9. Partner with Local Businesses: This is the same as having an event in the community, right?

10. Signage and Banners: These people don’t have yards to hold a yard-sign, and anyway it doesn’t provide the personal connection I want.

From this set of responses, I can see a possible decent strategy or two. So I wrote all of the above feedback into the chat, and asked it to come up with a few strategies that blend the methods I liked above. Here was the response: 

1. Text-to-Meet Strategy

2. Apartment Ambassador via Text

3. Geo-Targeted Digital Ads + Direct Text Follow-Up

4. “Virtual Door Knock” via Personalized Videos

5. Micro-events Leveraging Text Invitations

This got me interested. If you ignore the “Apartment Ambassador” and “geo-targeted digital ads” we’ve maybe got something here. Sending personalized video texts to folks in the building, telling them about an event we have upcoming in the area is a tactic I can see trying out, and might even work.

So you have to ask yourself: Who is managing this campaign — me or the AI? How well would this have worked if you didn’t have my experience to fall back on? Would the first time candidate even have recognized that apartment buildings will cause this sort of blockage, and which of these ideas would they have tried?

Campaigns are not just data with an organization stapled on to it. There are a lot of things that only a real person can do, like hiring staff, recruiting volunteers, refining messaging (though here AI can maybe lend a hand). 

AI definitely cannot handle any of these things on its own, and it can’t decide which things will even need to be handled. What it can do is support a skilled human campaign manager by acting as a brainstorming partner and as an assistant. Used correctly, AI makes strategists more efficient, although it’s a reasonable argument that I perhaps spent more time prompting than I would have if I had just come up with the “texting them” idea myself. While it can help good managers sharpen their work, it absolutely doesn’t replace them. 

To wit, if a campaign tries to offload too much onto AI, the result will be a mess. The content will sound robotic, the insights will lack depth, and the “strategy” will be disjointed because AI can’t possibly understand how all the moving parts fit together.

The worst mistake a campaign can make is assuming AI will do the thinking for them. It won’t.

Someday this might change. AI really is improving at a rapid pace and it’s definitely possible that by 2035, my answer will be different. 

But in 2025, it’s best left to the real humans with real working brains. For now, a human manager using AI privately to brainstorm, help with thought exercises, or a bit of coding help is the combination that’s going to win elections. A fully “AI led” campaign is one that will fail in comparison. Those that understand this distinction will be the ones that stay ahead. 

Caitlin Huxley bridges data and strategy to help moderates win close races, leveraging 15 years of experience. She’s the author of Ancient Wisdom for Modern Campaigns: Lessons from Sun Tzu’s Art of War.