Susan Pendergrass speaks with Todd Davidson, Vice President of Programs at the State Policy Network, about how artificial intelligence is reshaping the think tank world. They explore what AI is good at and where it falls short, how organizations like the Show-Me Institute can use it to become more productive without losing their edge, why face-to-face relationships will only become more valuable as AI-generated content floods the internet, how a Hawaii think tank used an AI agent to help fire victims submit legislative testimony, what good policy looks like in an AI-driven energy landscape, and more.
Episode Transcript
Susan Pendergrass (00:00) Great, well, thanks so much for joining us this morning. Todd Davidson of the State Policy Network, to talk about the topic du jour: artificial intelligence. Thanks so much for coming on to talk about it. I’m afraid to even say anything out loud about AI because by next week it’ll be…
Todd Davidson (00:11) Yeah, happy to be here. Thanks for having me.
Susan Pendergrass (00:18) Nothing really ages — it changes so fast. But I did just read that Mark Zuckerberg has an AI agent who is performing his CEO duties for him. Did you see that? Why not, right?
Todd Davidson (00:28) I saw that, yeah. And then he can just kick back, go down to his Hawaii bunker and just let Facebook run itself.
Susan Pendergrass (00:37) Yeah, I mean, I still haven’t really dabbled in agentic AI, but I know it’s right there and I’m going to want to do it soon. We’re going to talk about AI in the think tank world, but I have to check legislation and hearings and see how those things are going every day. I can well imagine an AI agent doing that for me.
Todd Davidson (01:01) Yeah, if it’s properly trained. So ShowMe Institute, to give the audience broader context, is a member of State Policy Network, and we have sister organizations like ShowMe in states across the country. The Libertas Institute, which is based out of Utah, did exactly what you’re talking about. Connor Boyack, the CEO, built a legislative tracking system that then feeds into their scorecard where they keep track of legislation. He said it took him about eight hours of work to code the agentic AI, but now it does the work automatically. Of course it needs fine-tuning and always has a final human observer that verifies everything, but it’s being used for those purposes right now across the country.
Susan Pendergrass (01:59) So we’re in the think tank world, and it’s probably more of an art than a science at the state level. Tracking the policies — first of all, thinking about the policies that we think would be best for Missouri, then doing a bunch of research on those policies, then creating content on those policies, then trying to talk to legislators and hope that they see our point of view, and that they enact actual laws that reflect those policies. That’s a really labor-intensive job. Which parts of that could you see being picked up by AI?
Todd Davidson (02:33) I’m by no means an expert on AI, but I work with someone who is. What has been explained to me is that AI is very good at synthesizing information. It’s very good at predicting — it essentially predicts the next word. It takes all these inputs and predicts the next set of words, which comes out to us as sentences. So if you are able to give it certain inputs — say, I want you to look at these bills, I want you to look at these things — and give it a sort of walled garden, it can then be prompted to produce any type of analysis that you want. The reason you want that walled garden is because AI can still hallucinate. It can make stuff up. Actually, this just went viral last week: a lawyer down in Georgia went before the Georgia Supreme Court and had AI produce her entire argument. It cited five fictitious cases, and the judges called that out. So you have to give it constraints and say, here are the data inputs, now summarize this for me. And it can get you a pretty solid first draft of that summary. Of course, you’re still going to need a human to go through and edit it and add voice and texture to it. But summarizing that data, saying tell me which of these align with our principles or does not align with our principles — it would be very good at that kind of thing. What it’s not going to be able to do is the creative part. When you think about what is the policy that we want to design for Missouri, what does Missouri need — it’s not at the stage where it could do that. That’s where you would still want Show-Me Institute experts to be crafting those kinds of things. But if something’s already out there and existing, you can summarize it and score it based on criteria pretty easily.
Susan Pendergrass (04:35) So given how quickly firms are moving towards AI — and in fact mandating AI because it’s such a time saver and productivity increase — how does a think tank position itself in that world? There’s so much talk about AI just replacing all of our jobs. Maybe it does replace my job — I don’t know. I’ve heard podcasts generated by AI in my voice, so it could be doing this job right now. I would like to think it wouldn’t be as great, but how does a think tank position itself? What’s our value add in that scenario?
Todd Davidson (05:12) Start by going back to what your mission and objective is. ShowMe Institute — and by the way, I am a resident of Missouri and a big fan of the Show-Me Institute, both from my SPN perspective and from my Missouri resident perspective — we have principles: free markets, a robust civil society, a thriving economy. We want the feds to get out of the way in a lot of cases. We want the government to get out of the way. And then how we execute that mission is through policy change, mostly at the state level, though I know you also work at the local level. So state and local policy change is the objective. How do we go about that? We produce research and then we advocate — in some cases talking directly to policymakers, communicating out to the public through op-eds and things so that the public then talks to lawmakers. And ultimately we get policies passed that lower the income tax, reduce barriers to work, and provide more options for kids in schools. So what AI is going to do is make research and content much easier to produce. By research, again, I mean that summarization kind of research — it’s going to make that kind of stuff extremely easy for folks to produce. Everybody’s going to have a research assistant. What AI cannot do is personal relationships. It will never be able to do that. What it also cannot do is tour the entire state of Missouri, know all of the history and relationships and connections of people throughout the state. So I believe Show-Me Institute and all of the affiliates across the country that are state and local based are going to have an advantage because you’re in your community. You know people, you know policymakers, you know community leaders, you know people that are affected by your policies. And that’s something AI is not going to be able to do. AI can look at the statistics and arguments and academic literature, and it could put together a brief, and that could be useful. It would make your job more efficient — you’d be able to produce those things in a fraction of the time you do right now. But then with that extra time, I would use it to go out and build stronger relationships in the communities, and then use those relationships towards policy change.
Susan Pendergrass (07:51) What about grassroots? More grassroots-type stuff?
Todd Davidson (07:55) Grassroots very much. AI is going to have an interesting relationship with grassroots. In one way, it actually makes it easier for grassroots individuals to engage their legislature. On the other hand, it’s going to create a flood of grassroots engagement digitally. So face-to-face grassroots engagement is going to have more impact. I’ll tell you a story: Hawaii had the terrible fire that destroyed Lahaina a few years back. Hawaii has terrible building codes — it’s incredibly hard to build homes there. That town was completely destroyed, so the state needed to relax its building codes in order for homes to be rebuilt. Well, they weren’t making this change. Show-Me’s sister think tank, called the Grassroots Institute of Hawaii, built an AI platform that allowed individuals to submit testimony to the legislature. Testimony has a higher bar, right? You can email your lawmakers pretty easily, but testimony goes into the legislative record and has to follow a certain format and be structured in a certain way. That’s not something that grassroots individuals were very equipped to provide. So a think tank would typically provide the testimony and then get grassroots supporters to send emails to lawmakers. What Grassroots Institute of Hawaii did was build an AI agent so that an individual could say, “Hey, my house was burnt down, I need these things,” and the AI agent would turn that into testimony and submit it directly to the legislature. It resulted in a skyrocketing number of testimonies being filed. Because of that, the legislature said, “Wow, we’ve heard from 500 constituents — we’ve never heard from that many constituents before.” So they relaxed their regulatory regime, and now homes are being built in Lahaina much faster.
Susan Pendergrass (09:48) Did they know that AI was doing it? Were legislators thinking, okay, this is AI?
Todd Davidson (10:12) That is why they went through testimony. Legislators’ email inboxes — they’re not reading their emails anymore, right? They get thousands of them. But through testimony, the AI was not making up the stories. The people had to fill out the content and explain their story. The AI was just structuring it in a way to make sure that it got submitted as testimony. I do think that is a bit of an arms race. At some point the same thing that has happened with email will happen — there will just be thousands of pieces of testimony and you won’t be able to read all of them. So there was a bit of a first-mover advantage. And once that becomes ubiquitous, I do think what you predicted is going to happen, where legislators just say, well, this is AI-facilitated. And that’s where it’s going to have to go back to face-to-face, bringing those people in.
Susan Pendergrass (11:08) I think you’re absolutely right. As more video content comes out and we all realize it’s AI — I just don’t really believe that any videos are real anymore. I don’t really believe pictures are real. I don’t really believe music is real. And it doesn’t necessarily bother me that much, but I think because of that skepticism and unwillingness to believe in digital content, things happening in real life right in front of us are going to take on higher and higher value, so that we know for sure that if I’m speaking to a legislator, it is me saying it and what’s coming out of my head. That’s about the only way we’re going to know if something is real — or the default is just going to become AI-generated.
Todd Davidson (12:01) 100%, I absolutely agree. And that’s where I think organizations like ShowMe are well positioned. Because you’re in the state of Missouri, you can be in Jefferson City or you can be in St. Louis or Kansas City in those face-to-face relationships. It’s going to make your government affairs personnel far more valuable, your fundraisers who can be face-to-face with donors far more valuable, grassroots activists that are face-to-face. It’s going to put a premium on face-to-face interactions for sure. I agree — there’s going to be so much content out there. You’re still going to need content because that gives you credibility, it gives you what you’re going to talk about. But then you’ve got to pair that with the face-to-face interaction, otherwise it’ll just get ignored.
Susan Pendergrass (12:47) And you can definitely see the gap when people are generating stuff through AI and they don’t know the subject matter enough — like you said about the attorney. But there is definitely a role for humans to say, I mean, I do this all the time with AI: I’ll say give me five of these things, give me five infographics or something like that. But the human has to know which one is the best or which one makes the most compelling argument. AI simply really can’t do that. So while some people would love to believe that AI is going to run the world, I do believe there is an emerging role for human discernment to know which AI products are better than other AI products. Would you agree with that?
Todd Davidson (13:32) Yeah, 100%. I think the sweet spot is utilizing AI to make yourself more efficient or do things that you don’t like doing. But then that raises you up into that discernment phase where you’re the one making the call. I do this all the time — I’m having conversations with AI to increase the outputs. I should not spend any time making infographics. I’m not good at it. But I can have a conversation with AI where it produces that infographic much more effectively than I could. I’ve also found that, if you put the prompting on it, it can help you find those particular sources that you’re looking for. Say you want to write a survey on school choice research — it can help you gather all of those materials much faster. But then you have to make sure that it’s of high quality.
Susan Pendergrass (14:35) What do you think about the current pushback on AI-generated pictures? Do you think that is just a learning phase we all need to get through? Some top artists on Spotify have been determined to be AI-generated.
Todd Davidson (14:57) Really?
Susan Pendergrass (14:59) Yeah. The number two Christian artist is just AI, and across all genres there are artists with millions of subscribers who are just AI-generated music based on what AI knows we all like. So we do like it. Does it matter that there’s no real person writing the music? I don’t know.
Todd Davidson (15:12) It’s kind of sad. Yeah.
Susan Pendergrass (15:21) I know the initial reaction is, that’s sad. But then after a while you’re like, I don’t know.
Todd Davidson (15:26) There is going to be intense pushback to all things AI. AI is very unpopular right now. I saw some polling just last week that showed it is the number one concern of voters. There will be a populist pushback against AI. We’re seeing this pushback against the data centers. There’s even polling that showed a plurality of the population believes it’s immoral to use AI. And I think it gets at the core of some of what you’re talking about here — yes, there’s this very popular, satisfying music, but it loses some human element because there’s not a human behind it. I do think we’re going to see a lot of pushback to AI on multiple dimensions. There’s that cultural dimension. There’s the economic anxiety dimension right now: a fear that AI is driving up energy costs, a fear that AI could take my job. There’s going to be pretty significant pushback. Right now we’re mostly seeing that in anti-data center efforts, trying to stop the building up of data centers across the country. I was looking at some Democratic pollsters today who were pitching that Democrats should advocate for a guaranteed job, guaranteed income, guaranteed healthcare, and a guaranteed home if you lose your job to AI. That kind of populist messaging is going to resonate with a lot of the public. What is the response going to be to that? What are the other solutions that we could advocate for that both allow the continued growth and opportunity and also allow continued innovation around AI, because we’re going to need AI to continue to develop?
Susan Pendergrass (17:30) It’s already here. I mean, we’re doing this in reverse order. And I think my opinion is that massive new technologies always get pushback — like the car. People were on their horses, and then we started designing roads for cars. Calculators got a lot of pushback, the internet got a lot of pushback. But ultimately people decided that they liked it better. I think AI is the same — we just have to figure out how to work with it. And I know that it is threatening to take a lot of jobs, but I see it more as a good thing. It gives us an opportunity to become the expert over AI. AI is not going to be the expert — we still need the human component. Like you said, face-to-face interactions. Legislators are still going to know what Missourians want and how to represent their constituents, and those are real-world issues. The data center pushback is because I don’t want to look out my window and hear a buzz and see a data center — I don’t want all that land going to data centers. That’s a real-world, in-person issue. But I just think we’re going to have to learn to work with it. I don’t think robots are going to — maybe this is where I don’t want to say things out loud — but maybe the robots will take over the world, I don’t know. But personally I feel like it is helpful to get a lot more content out, because you don’t know what’s going to resonate with stakeholders. Whether it’s a video or an infographic or a report or a different type of content, the fact that we can generate these things much more quickly I think is a benefit to us, and it makes the in-person time more meaningful to me.
Todd Davidson (19:11) You’re absolutely right. When a new technology comes out there’s going to be pushback, and organizations like ours have to figure out what’s the policy framework that allows that innovation to thrive without getting in the way. And fortunately we have a lot of those policies already. Like Avery, your colleague at Show-Me Institute, talks a lot about energy. One of the biggest pushbacks on AI is that it’s driving up energy costs. There’s some research that shows that’s not quite what’s happening. What’s happening is a lot of green policies that got passed in the 2010s are coming to roost — the renewable portfolio standards and those things are really what’s driving up energy costs. But even still, what can we do to make energy more affordable and reliable, even with a bunch of data centers added to the grid? And Avery’s got good policy on this: expanding nuclear power, expanding the use of reliable energy sources.
Susan Pendergrass (20:23) It’s separating out consumer electricity from data center electricity. You can carve these things differently.
Todd Davidson (20:29) Yeah, that’s another one — where the data center has its own power source. So there are policies out there that can mitigate it. And on the job question, unfortunately AI is happening at the same time that we’re having a continued cost of living and inflation issue. It’s one more thing that is driving anxiety. It’s not the root cause of what’s going on — we’ve got other factors that we need to address to get inflation under control, particularly on the energy side.
Susan Pendergrass (21:08) Yeah, but I do think it’s great that we have so many opportunities to expand or improve how we do things. In our little corner of the world, which is think tanks, we’ve been doing things kind of the same way for a long time. So I think a new approach to how we do business is a welcome change, and I think we could be a lot more effective.
Todd Davidson (21:38) Yeah, I think we’re going to see far more productive think tanks on the research side. On the litigation side, I was talking to Wisconsin Institute for Law and Liberty. They litigate a lot of cases. With the advent of AI, every lawyer essentially got a legal clerk right away. They went from nine lawyers and a handful of legal clerks to nine lawyers who each now have their own AI legal clerk. It’s dramatically expanded the number of cases they can take on. And the same thing on the research side. On the marketing side, production of content is going to be quite a bit easier and more cost effective as well.
Susan Pendergrass (22:26) Well, I appreciate having a chance to talk to somebody who has a positive perspective on it, because I do hear a lot of doom and gloom when it comes to AI. I was reminded by somebody that many of the scenarios in movies and books about AI are very dystopian, but perhaps it’ll be utopian. We don’t know. It’s all in how we approach it, right?
Todd Davidson (22:48) Yeah, it is. It’s going to be an exciting new world that we live in and we’re right on the frontier.
Susan Pendergrass (22:54) Anyone with little kids, like you — who knows what the world’s going to look like when they’re going to college. So you’ve got to stay flexible, right? Well, thanks so much, Todd. I appreciate you coming and talking to us about it. We’ll have to talk about it again sometime soon when the whole thing has changed.
Todd Davidson (23:02) Yep, stay flexible and always be learning. Yeah, sounds good. Thanks, Susan.
Produced by Show-Me Opportunity
