Maya Hutchinson (00:00):
If you really need to get a campaign or an ad out on a super local level and you're running for city council that previously wasn't available to you. And so that's what AI is enabling.
Eric Wilson (00:14):
Welcome to the Campaign Trend podcast where you are joining in on a conversation with the entrepreneurs, operatives, and experts who make professional politics happen. I'm your host, Eric Wilson. Our guest today is Maya Hutchinson, the founder and CEO of Battleground ai, an integrated advertising platform built for politics effortlessly combining data messaging and ad creation in one place. Maya is a data-driven marketer with roots in democratic politics. In this episode, we discussed the tangible ways campaigns are using AI and how they can more fully tap into this emerging technology. Now, Maya, I'm going to link to your campaigns and elections posts, which really got me thinking about this conversation in our show notes. But you mentioned that campaigners underutilized AI in 2024. Could you share some specific examples of what those missed opportunities were and where should we focus going forward?
Maya Hutchinson (01:07):
Yeah, absolutely. I think we saw a lot of headwinds while this was still an early technology, the chat GBT was on the scene in 2022, but 2024 was the first election where folks were really starting to figure out and test out AI and more personal opportunity. But I think given that there was a lot of fear around that new technology and a lot of risk involved potentially, or that was the narrative,
(01:41):
Folks shied away from more publicly adopting. In fact, a lot of organizations said they didn't want to have anything to do with it or not use it at all. And I think that can be really harmful, especially when it's the advent of a brand new massive change to the industry. We need to start small and experiment and do that in ways that we can test and learn, but we need to actually to start to do that. I think areas where we saw opportunity, our platform was focused on the content creation and advertising side, and we had some really great small test cases that I think given maybe we were able started late in the cycle, but if it was even a little bit earlier, having more of those, having especially down ballot folks who may be under-resourced and always strapped for time, really be able to do more with less when it comes to content, when it comes to ads, running ads, you can do that with smaller budgets. You can do a lot more now with the help of AI and avoid any of the pitfalls or risks or fear of miss or disinformation. So I think when it comes to the content side, that was certainly a potential missed opportunity for a lot of organizations who could have benefited.
Eric Wilson (03:04):
You probably had this experience too with reporters who were reaching out last cycle, seeing how are people using ai? And I would always tell 'em it's in the most boring and mundane ways you can imagine. They all wanted to write a story about how Terminator is going to take over elections, but as you point out, it's pretty sedate and kind of important, but behind the scenes things. And so how would you categorize the different hurdles holding campaigns back from really fully embracing AI tools?
Maya Hutchinson (03:32):
Well, I think there's a lot of uncertainty, right? There was, you're in the midst of election, so that breeds a lot of anxiety and uncertainty naturally, of course. And then also there's lots of pending state legislation that was happening at the same time. There also, there were these small one-off cases that were getting attention around the misuse of ai, and they were very, very small and they were caught. And I think there were examples. There's been that fear and misinformation. I think we talked since the printing press, right? It's not because of AI that that's happening. And so I think those are really where that uncertainty came in for people and then they're going to stick to what they know. And I think in campaigns you often do that because of the short windows. You just stick to the same playbook over.
Eric Wilson (04:32):
And that's certainly our tendency. And it also started with the platforms sort of saying, look, we are just not going to go through this headache. You're not allowed to do political stuff on the platforms. And so that put kind of a damper on people's ability to explore it even. And I think that was kind of shortsighted and throwing the baby out with the bath water, if you will. And one of the big challenges is just getting people to try new things and you don't want to poke your head up and be criticized or accused of committing disinformation. But as I've pointed out in previous conversations we've had, this stuff is already illegal. The example of the voter suppression call in New Hampshire with the New Orleans based magician was the culprit there. What he did was illegal, whether or not he used ai. And so it certainly had kind of a chilling effect. And so battleground AI is working in an area where I think we've been comfortable with AI and machine learning for a while. So this is the example I go to of, well, if you've ever looked at someone's Facebook ad library archive, the reason you see dozens of variations of the same creative is because they're using ai. And so we're really comfortable with that for modeling and optimization. What's the advancements we're seeing with generative AI relative to this media buying and optimization?
Maya Hutchinson (05:58):
And I would say I think a lot of people are still uncomfortable with a lot of it, but we're getting better. And I think certainly from folks on the targeting and data side, there is that element that folks have been doing. And then I would say when it comes to battleground ai, the opportunity that I've saw from working in advertising, especially when it comes to politics, is that deeper integration across the full life cycle of your ads from your polling and research all the way through to actually the data on the ad platforms itself. These things have to be better connected and better informed. And while we as humans have the capacity to do that, it's not efficient.
Eric Wilson (06:47):
And we don't always have the time.
Maya Hutchinson (06:49):
We definitely don't have the time when it comes to campaigns. But even if you're working in sort of a long tail political organization, that's really hard to do as individuals. It wasn't possible until now. This technology is unlocking a lot of opportunity and also making this more scalable. So I think when you're working on a presidential campaign or a giant Fortune 500 company and you have unlimited budgets, that maybe isn't as pressing, but if you really need to get a campaign or an ads out on a super local level and you're running for city council or school board, that previously wasn't available to you. And so really trying to make these tools and these resources more scalable for folks, and that's also what AI is enabling.
Eric Wilson (07:41):
You unlocked a really key insight for me just now, which is that we talk a lot about the siloing of data within a campaign, and usually that means digital might not be talking to field or finance and field aren't talking, and maybe the TV buyers aren't talking to the streaming buyer, all kinds of different permutations of how that might happen. But I hadn't thought about it in the context of within your own silo, you are generating data and say, how can we make ourselves smarter with that? And yeah, it does take time. And so that's really interesting. I always go back to this criticism of I really don't like the way I quibble with the word dashboard for advertising because if it were like a car dashboard, you would crash into a wall if you were looking at it because it doesn't give you that quick information. And so now with ai, we have something that can process all of that information. That's really interesting.
Maya Hutchinson (08:34):
Yeah, there's definitely internal silos that we can start to break apart and and I think also when it comes to sort of the individual, the data that we're able to better utilize not just voter data or not just that comes to unstructured data that can be very informative that we weren't able to fully incorporate in an effective way before.
Eric Wilson (09:01):
And one of the things that you point out is that despite the fact that national issues continue to dominate the conversation, even at the local level, you argue that local research is important. And so help us understand how AI can localize campaigns more effectively as we try and get away from the nationalization of state and local politics.
Maya Hutchinson (09:24):
Yeah, I'm obsessed with politics is local idea because I really do think that especially when you look at national polls, that's a very small sample size. And yes, you can extrapolate it if you're running for a national office, but again, if you're a local candidate, even running a poll in a small district is hard if it's not big enough or you don't have enough resources and the information that you can start to gain from AI and just experimenting with it. But OpenAI has launched their deep research tool and it's absolutely incredible. And so I think that in and of itself is a game changer. You can test that out for $19 and get really interesting, and I am sure you too experimented and asked it about what's happening in a certain race in a certain state, and the level of information is eyeopening, honestly, and how fast it works.
(10:22):
And so I think those are things that kind of unlock ideas and opportunities, especially if you work in a field that needs that kind of information and needs it rapidly. And so I think you can really start to focus locally, keep it up to date, listen and constantly understand what are people talking about? What are they talking about online, what's news and media talking about locally? And really being able to fine tune that conversation with people that was previously really hard to do. And it took a lot of time, took a team of people, and it's not that you still don't need a team of people, but it's like, Hey, we can do this a lot faster and win races.
Eric Wilson (11:02):
You're listening to the Campaign Trend podcast. I'm speaking with Maya Hutchinson from Battleground AI about some practical use cases for AI and what campaigns need to do. Now. We can't talk about AI without discussing the ethics question, particularly when it comes to AI generated creative. So where are you drawing the boundaries in your own mind right now?
Maya Hutchinson (11:25):
Yeah, I think the things that I think about all the time is the idea of trust, the idea of truthfulness, which we talked about, and then of course at the most important, but also kind of bare minimum to working in this space is the laws and regulations around it, because there's obviously a lot of work to do there, but it's slow and not moving as quickly as we may have anticipated. And so I think when it comes to the platform and it comes to our work, we try really hard to be as transparent and communicate what we're doing as much as possible. We're a self-service platform for left leaning candidates, and so we're not controlling every single element, but the way we structured it is transparent, what models you're using, how and where your data is being used. And we have structured it in a way that we believe versus a free flowing chat.
(12:25):
We really try to have a structured, prompt engineered in the platform so that you're getting a consistent output. While prompt engineering is an important skill, it is hard, and not all folks who are going to be working in any space are going to be experts at that and get the same thing every time. And we really want to make sure that people are producing the right content, factual information in a really transparent way. And that's what we can do as a company to set those standards and to work with people to understand that while we can't set the laws and we can't handhold every single person, we can at least set our own structure around that usage.
Eric Wilson (13:07):
I think that there's also this, I don't know, conceit of, we need to put it right in front of people's faces and let them know that this is ai, otherwise they're going to be confused or misled. And I think probably where we're going to end up is more like in the way that you mention of if you want the information about how this is being used in this product, you can find it out. So I think about digital cameras. I don't know what any of the metadata that's attached to the photographs that I take are what's the difference between different compression algorithms and codex and things. Someone way smarter than me understands that, and I could go to them and say, okay, well, you really want to use this kind of camera doing a better job of lossless compression, whatever it is. To your point, we don't have to become the experts in the AI to know what's going on, but again, it goes back to we have a duty and an obligation to be moral and ethical and follow the laws. How concerned are you about these lawmakers at the state level who are starting to make moves into trying to regulate ai, especially as it relates to political?
Maya Hutchinson (14:22):
Yeah. We have to put our trust into the folks that we fairly elect to office, even if we don't fully agree with everything that they're doing, and we make those decisions to elect those folks because we think that they're going to engage the right stakeholders in that decision making process. I don't think that every elected official is going to know and be an expert in ai. That would be, they should be informed, but the rest of us, they should seek out information. They should talk to companies. They should talk to nonprofits and other organizations and researchers and experts in AI field to figure out what's going to make sense, how do we get ahead of this technology? But also, we're already late on this technology. These algorithms and large language models have already been developed. They've been working on open AI since 2015. Their legislation didn't pass.
(15:27):
I guess Colorado and California have some of the biggest ones, but that didn't pass until last year. And so I think we're already behind. And so I think it's really challenging. To your point before, this is sort of going to be a foundational element of every technology that's built moving forward. And when you were talking, I was thinking it's kind of like nutrition facts. You're going to eat something and if you really want to know what's inside of it, you can figure it out. I think to that point, yes, we should know what I think a lot of models are. They're now linking to the new sources and they partnerships, and they're developing more of a structure around where they're getting that information and how they're built. And obviously we saw that with deep seek and having many more models, but I think, yeah, we have to educate them.
(16:21):
And also, it's going to be really difficult to have piecemeal state by state legislation for something that touches pretty much every industry. This isn't transportation and ride sharing. That's one sector. AI on a state by state legislation level is going to be really difficult and complicated for any organization to manage. It's going to be costly and restrictive and challenging for businesses to start up and function across the country because very few, I mean obviously in politics we're state by state in many cases, but most folks don't. Don't think like that. We are all operating in the United States, so we need to think bigger picture about what our global policy is and also how do we work with other countries,
Eric Wilson (17:09):
Right? Yeah. That's
Maya Hutchinson (17:11):
A whole other, yeah. Yeah, and so I think we do need to talk about it, but I think it's going to be challenging.
Eric Wilson (17:18):
Yeah, Europe was very quick to regulate this, and I think the regulation we have to say comes from a good place where people say, Hey, this thing is coming. It is concerning. It could cause problems, so let's get out in front of it. And the risk, of course, is that you hamper innovation. You potentially cut out some good uses of generative ai. An example I always go to is DeepFakes could be used for good purposes where if we're trying to translate into individual languages, that's particularly helpful. Are there any other examples that you might point to where it's like, Hey, if we just say, Hey, all AI is bad, we're going to be missing the boat here?
Maya Hutchinson (18:01):
Yeah, I think there are a lot of, I was actually trying to think of the, I think it was either in Venezuela perhaps where reporters were using AI replicas of themselves to report out of fear of retribution about what was going on.
Eric Wilson (18:16):
Interesting.
Maya Hutchinson (18:18):
I do now need to get my own sources on that too, but I do think that there are opportunities to your point on good uses around how do we take not just DeepFakes, but opportunities to expand knowledge, especially when it comes to freedom of press or many other opportunities in sectors that I think maybe aren't always first to adopt new technology, especially nonprofit sectors where you're constantly strapped for time and resources and now having this unlock of being able to finally, I can do more with this. I have what is almost essentially a very well-trained intern on my team, and so we were talking about that this week too. Being able to have those folks, not literally, but AI on your team to help you is a big unlock for a lot of organizations, and I think we will start to see, I think as folks do have more opportunity, I think we're getting into a space, I think on that sort of tech adoption curve where we start on the innovators, but we're slowly going up the curve to more and
Eric Wilson (19:26):
More people crossing the chasm.
Maya Hutchinson (19:27):
Yeah, exactly. To more and more, and I will say on the regulation piece too, I was at an event in Europe, and while they do have that regulation place, it actually seemed to me that was starting to spur more people to think about it. They had a big dialogue about what that kind of framework was, and of course you can have a policy debate about whether it's right or wrong and whatever, but there was a big appetite for innovation and so many people doing that, and so I think there's opportunity there that it doesn't need to always be a restrictive element. We can work with corporations that also want structure around those things, so I see opportunity there, but yeah, there's what, 400 million people just using chat GPT alone, adoption is fast,
Eric Wilson (20:17):
And so we've been focused on how campaigns can leverage AI to reach voters, but one of the things I'm increasingly thinking about is how are voters going to be using AI to engage with or scrutinize or even block campaigns? What's your prediction on that front before we wrap up
Maya Hutchinson (20:40):
When it comes to text messages? I would love for AI to be able to mitigate those. I think what is more and more happening is that, well, one, yes, voters are, people are using this all the time, and I think more and more it's starting to take a big share from Google search in that you need to think about how you are appearing as a candidate, as an organization, as a company. You're going to show up when you ask questions, when people are asking questions, what AI tool should I use for advertising? I would like to show up. It's like you want to show up in those things and you want to show up well, so really thinking about that is a new whole new platform and interface where you need to make sure, and that comes down to media and content, but making sure that you're addressing that with voters because they're going to be using this all the time. I don't know as much from a suppression standpoint, I think, but much more from an informational standpoint and tracking what folks are doing. For sure. I think especially as you start to get into many primaries and races that are coming up, it's definitely something that should be on a lot of consultants' mind.
Eric Wilson (21:53):
I don't have one. We need to figure out a way to talk about chat, G-P-T-S-E-O or search engine optimization. Don't know if we have a term yet for that, but I don't even think there's a reliable playbook to making sure that in the way of like, okay, we know the rules of on page SEO and off page SEO for Google. It's such a brand new and ever changing field. We don't know. I can't advise people on best practices to appear on Chad GPT.
Maya Hutchinson (22:25):
I think they're going to, my prediction from last year was that, I mean, perplexity has already announced it, but they're going to have ads. That's going to certainly be one format that here we go. Sure. Your kind of place at the top, much like Google search, there will be ads within, I don't know every platform, but certainly perplexes is kind of the first one I believe that's been pioneering that model, and so I think ads will be one format for sure. But yes, just like SEO and sort of unlocking that secret sauce, which has been very successful over my people, there would be another secret sauce to learning and educating models on your company, your candidate, your individual self. So yeah, you thinking about it from a lot of angles,
Eric Wilson (23:13):
It's great. I mean, you're probably seeing this, but you started to see referral traffic from Chad GPT or Claude, and
Maya Hutchinson (23:19):
I don't
Eric Wilson (23:19):
Know how to make that happen, but it's
Maya Hutchinson (23:21):
Happening. Yeah, I posted that on LinkedIn. I was really shocked. As I'm looking at my Google Analytics, I'm like, oh, chat GBT is actually for the first time its own referral source up there in, I was like, that's pretty wild. I had not seen that before, and I have been looking at Google Analytics reports on a lot of websites for a long time.
Eric Wilson (23:44):
Yeah, it reminds of back in the day when you'd see, oh, 12 people came to your website from the Xbox browser or something, it starts to crop
Maya Hutchinson (23:52):
Up. You're like, what is that?
Eric Wilson (23:53):
Well, my thanks to Maya for a great conversation. You can learn more about her in our show notes and we've got a link to her company Battleground ai. This episode made you a little bit smarter, and I certainly gave me some new insights. All we ask is that you share it with a friend or colleague. You look smarter in the process and people learn about the show. It's a win-win all around. Remember to subscribe to the Campaign Trend podcast wherever you get podcasts, so you never miss an episode, and you can visit our website@campaigntrend.com for even more. With that, I'll say thanks for listening. See you next time. The Campaign Trend Podcast is produced by Advocacy Content Kitchen, a media production studio.