How Could Artificial Intelligence Change the World of Think Tanks?, with Dan Baer

Notable Quotes

• The utility of research from think tanks evolves: “Much of what we have to think through is how we navigate a world of content surplus, because most of our organizations were created for a world in which there was a content shortage.” (5:15)
• Growth mentality for artificial intelligence: “One of the things that I repeat to my colleagues in my capacity leading our research is this is not about policing; this is primarily about learning together and making sure that we are appropriately acknowledging the ways that we’re using new tools in our work.” (17:41)
• Human intermediaries: “The successful expert or scholar will be responsible for mediating the intersection between knowledge or information and structures of power … Some people will be content to swim in the AI slop, and others will want a life raft. I think there’s an opening for human experts in that world.” (28:05)

Top of the Stack

• “Mission over Model: What Might AI Mean for an Organization Like Carnegie?” by Dan Baer
• The End of History and the Last Man by Francis Fukuyama
• The Wonderful World That Almost Was: A Life of Peter Hujar and Paul Thek by Andrew Durbin
• “Terra Incognita: The Economics of a Shrinking World” by Jesús Fernández-Villaverde and Patrick Norrick

The Full Transcript

Kristin Hayes: Hello, and welcome to Resources Radio, a weekly podcast from Resources for the Future (RFF). I’m your host, Kristin Hayes. My guest today is Daniel Baer, who’s the senior vice president at the Carnegie Endowment for International Peace, which I’ll just call Carnegie for shorthand today. Back in August, Dan posted an article on Carnegie’s website entitled “Mission Over Model: What Might AI Mean for an Organization Like Carnegie?”

I’m going to offer what is some very sincere flattery and note that I really love this piece. I love its humility. I love the beauty of its writing, which I personally think matters even more in a world where we have a lot of outputs from large language models (LLMs). And when this piece came across my desk, I basically immediately reached out to Carnegie to see if Dan would be willing to talk with me about it.

The piece offers five what Dan refers to as “emergent intuitions” for how AI [artificial intelligence] will affect organizations like Carnegie and RFF moving forward. And I would argue it offers wisdom for anyone who thinks about knowledge creation, expertise, communication, and relationships in the age of AI. So, while this is a little bit of a departure from our usual Resources Radio fare, I think our listeners are in for a real treat. So, stay with us.

Hi, Dan, and welcome to the show. I’m super excited about this conversation and really grateful for your time.

Dan Baer: I’m thrilled to be with you.

Kristin Hayes: We always do like to introduce our guests, and I’m wondering if you can say just a little bit more about yourself, your areas of professional focus, and more about you.

Dan Baer: Sure. I’m from Colorado, originally, and live in Washington, DC, now. I’ve had a number of past lives, including as a cowboy, an academic, a diplomat. And I’ve been at Carnegie for the last six years.

Kristin Hayes: Fantastic. Okay. That’s quite a combo. I could ask you a million things about being a cowboy and being a diplomat. Those frankly sound like fascinating podcast episodes in and of themselves.

But I do want to spend time on your piece about AI today because, again, I really enjoyed it. I’d love to talk through those five emergent intuitions in turn, if that works for you.

The first one of those reads, “We might produce less, not more.” So, obviously I live in a think-tank world. It does feel like there’s some pressure to produce more and faster. And you definitely acknowledge from the outset that that intuition might seem counterintuitive, but you also make a really compelling case that we might actually publish fewer, but more genuinely novel, pieces of research. Again, I’m quoting you!

So, start there. Talk to me about how you see the volume and the types of research that organizations like ours produce. How do you see that changing?

Dan Baer: First of all, let me say at the outset, you correctly characterized them as emergent intuitions. The whole point of the piece overall was not to make any bold predictions or prescriptions for what think tanks should do. Instead, I felt like a number of the people that I know in the knowledge economy space were all grappling with big questions about how AI might change the way we do what we do and how we accomplish our missions. And even though it feels scary—particularly for those of us who are in the business of making scholarly predictions or prognostications—it feels scary to say something about something that you have low confidence in. It felt important to me to try to catalyze a conversation. So, anything that I say today or that I wrote then was really not meant to be a final word, but rather, I hope, helpfully a first few words that really consolidate some conversations that I’ve had over the last few years with people inside of Carnegie and outside. So: caveat, caveat, caveat.

But I think one of the fundamental shifts that is behind maybe all five of the intuitions in different ways is just this sense that we are quite significantly moving from a world of information shortage to a world of information surplus, if we haven’t already. As I say in the piece, the internet in many ways democratized the distribution of information, and obviously that has had enormous direct and knock-on effects. AI really already has started to and will continue to democratize the creation of content.

I think that much of what we have to think through is how we navigate a world of content surplus, because most of our organizations were created for a world in which there was a content shortage.

Now, I understand that there’s not tens of millions of people waking up each morning and looking for what RFF has published today or Carnegie has published today, but there has historically been a core audience of people, some from the general public, some obviously in government or in other policy spaces, that have looked to us to fill a gap that it just wasn’t possible to actually have a comparison of land use approaches in country X and country Y at your fingertips, and you could go to RFF or some other think tank and reliably find something close to what you were looking for.

The advent of LLMs means that many of those things that people used to look up for baseline knowledge—they no longer need to come to us, at least for some kind of ersatz version of it, which may be good enough. And I think we have to remind ourselves that, for many uses, 80 percent as good as what we can do is good enough for helping the person accomplish what they need to accomplish.

It just seems to me that in a world of information surplus, we may need to focus more on the truly high-quality, novel research—the stuff that isn’t out there that can’t be regurgitated through an LLM yet. And that is where we’ll end up focusing our time, not just because we want to have a comparative advantage as humans over the machine, but because actually, if we’re trying to reach a certain policymaker or set of policymakers, the thing that we can offer that will be compelling to them to still come to us is something that is new and novel.

And that kind of research actually takes more time than what I call “explainer research” or just purely synthesizing research. Those products still have value, although I think they are the ones that are most likely to be first replaced by (or some of them will be replaced by) AI.

But I think over the medium term, we’re going to see more of a focus on what I call genuinely novel, and that takes more time to create. So we might end up publishing fewer words in hard publications or even on our website, but they may be better words on average than the ones that we’re publishing now. I offered that as a counterintuitive one, because I’ve already seen funders and others who think, “Hey, AI is here and it should really accelerate the research production pipeline.” And I wanted to challenge that assumption and think it through.

Kristin Hayes: Mm-hmm. Yeah. I find this fascinating. Just thinking about RFF’s own product types, we have a range of those. We do have explainers. We have what I would consider to be excellent, but nonetheless fairly what we refer to as “101 content.” And those are among the most downloaded products on RFF’s website. There’s a huge demand for those right now relative to some of the deeper, thicker, juicier, and maybe arguably more novel pieces of research. Those explainers have had a real demand signal, if I can call it that. I think it is a little bit existential to think about, well, what if that’s not our role anymore?

But I also think there’s a world where, as AI output is there but not necessarily fully trusted, people will still come to a place like RFF or Carnegie for even that baseline information, simply because it’s hard to know what else to trust.

I know we’re going to have a chance to talk more about expertise and the value of expertise later, but I’m just curious if you have any additional reflections on, Is there still going to be a role for us as experts, even in that baseline information provision?

Dan Baer: There may be. To the extent that the piece has been successful so far, it’s been in generating the kinds of conversations I wanted to have. And some of my colleagues have said, “Isn’t the answer here that we should actually be writing many more words and trying to get good information into the answers that people are getting from LLMs?”

You mentioned that your explainers are highly trafficked. I assume the same general correlation applies for our work. These days, I’m sure your web people are telling you more and more of that traffic is not human beings. It’s bots that are combing our websites and harvesting that information. And it doesn’t surprise me at all that LLMs themselves find the explainers the most useful for them to regurgitate or to build an answer around.

We may actually just be increasingly speaking to the world through LLMs, which—we don’t monetize our content that way, so from an organizational economics standpoint, that’s not a threat. And you could argue, maybe our content is actually reaching even broader audiences than it was 10 years ago, because everybody who’s using LLMs to access it wasn’t necessarily coming to our website. So maybe it’s actually good for getting good ideas out there.

But I do think that the explainer as a form is harder for us to say is something that we uniquely can do in a world of LLMs. And I think eventually that’s going to cause us to ask questions about, What can we uniquely do?

Kristin Hayes: Yeah. That is going to be the challenge, for sure.

And that’s actually a great lead-in to my second question, which is about your second intuition, which reads, “As researchers, our core skills will be curation, refinement, and creative improvement of ideas, often in collaboration with others.”

So again, I’m sorry to be all goopy about this, but I’ve really loved your writing here, too. That’s just a beautiful sentence, and it says a lot in those words. But I guess my first question for you on this is, How does that actually reflect an evolution of the skill sets that researchers currently deploy? In other words, what’s going to potentially change about the role of a skilled researcher moving forward?

Dan Baer: Obviously, I’m sure that RFF, as Carnegie does—we hope that researchers have those skills now and that they use those skills now. I think that this is less about some new skill set and more about the emphasis of those particular skills among a researcher’s skills. I guess it comes back to the idea that it’s going to be harder and harder to distinguish ourselves simply on the ability to write in a way that synthesizes or summarizes existing ideas and that we will have to be able to demonstrate—maybe not literally, but we will have to be able to create written product that is either making ideas better, or making ideas “pop.” And there, I come back to the idea of going from information shortage to surplus. In a condition of surplus, making good ideas stand out becomes part of what we do.

And I guess the contrast that I try to make … frankly, this is one that’s uncomfortable for me. Because I make the point that there was a time when a researcher at Carnegie could get by (and when I say “get by,” I don’t mean sneak by, but could be adding value for an entire career) merely by being an excellent writer. And that they were adding value by translating ideas into prose, which was something that was a uniquely human task. And I think that that is going to become less and less possible.

In addition to being a good writer, or even in some cases more important than being a good writer, is going to be either being somebody who can really refine and take to the next level a set of ideas or who can really, for lack of a better term, “market” those ideas to the world in a way that makes good ideas stand out.

Kristin Hayes: And I’m curious if you have a vision for what that “popping” looks like. Do you have any examples of where you’ve seen that happening already? Researchers have been focused mostly on clear communication and all that good stuff, but what does that next level of evolution actually look like?

Dan Baer: I think you could probably point to a number of things. We can have a long conversation about the evolution of the ways in which people are consuming ideas. One of the conversations that we’re having (as I know many other formerly staid knowledge-creation organizations are having) is, How do we integrate vertical video into our outreach? How do we translate scholarly outputs into things that people can consume in 90 seconds? That’s one version of marketing ideas, I guess.

But I would also point out that that itself is not a new thing. Many people in the international relations field still today talk about Frank Fukuyama’s book from 35 years ago, [The End of History and the Last Man]. The idea of the end of history is a frame, or a heuristic, that has ended up being an extremely powerful way of organizing a set of ideas that have percolated through a number of discussions in international relations.

I think what is special about that (and you could point to other examples) is being able to come up with killer heuristics that help people organize, again, information surplus and make sense of complex ideas in a way that is digestible. That doesn’t require that you put it on TikTok or in a reel. Those stunningly simple articulations of complex ideas may come to somebody in an aha moment, but they usually don’t come in the first 90 seconds of thinking about something. They are the product of years of reflection and expertise that allows you to distill something down. And I think that is part of making ideas pop.

Kristin Hayes: Those are great examples.

Let’s talk about writing for a second. This is something we talk about regularly within RFF, and I want to quote another line from the piece that I really loved, which reads, “Indeed, for most of us, the act of writing is the most important ongoing training we have for wrestling with ideas.” So, really the sense that writing is not just about stringing words together; it is not just about, in LLM speak, predicting the next word that’s coming. It’s really a much more intellectually engaged and engaging process.

I think that sentiment really informed the development of our AI policy within RFF. How do we make sure that we’re not losing the writing and, with it, some of that wrestling with ideas?

Actually, let me just take this opportunity to ask: Does Carnegie have an AI policy? And I’m curious if so, what boundaries does it place around AI-assisted or AI-generated writing?

Dan Baer: We do have an AI policy. We’re on version 3.0 of our AI policy. It’s evolved. Obviously, the technology is evolving, as well as people’s use of it, so it has evolved. But we rolled out our most recent version of it actually last month.

The top line is that we don’t allow AI-generated content to be on our platforms. We do articulate (and it’s two pages with a bunch of FAQs, so I can’t encapsulate it all here) and highlight some uses that we think are really good for people to be experimenting with, as long as they document it and acknowledge it.

I think, more important than the particular lines that it tries to draw (because I think we’re all learning as we go here), one of the things that we’ve tried to do is to foster an evolving culture around AI use that is focused on transparency and not being embarrassed and making it normal to have conversations with colleagues about how you used AI in the course of certain research.

So, what we’re really trying to do is facilitate a culture of communication around this, so that when I’ve used AI to help me pull together a data set and to do some initial analysis, I can share that with the people who are peer-reviewing it, or with my reviewer or the publications team, and say, “Here’s how I used it.” And then we can talk together about how to stress-test it, how to acknowledge that use, how to document a methodology if appropriate, et cetera.

One of the things that I repeat to my colleagues in my capacity leading our research is this is not about policing; this is primarily about learning together and making sure that we are appropriately acknowledging the ways that we’re using new tools in our work.

Kristin Hayes: That sounds very similar, I think, to the way that we’ve approached this at RFF: emphasis on transparency, making sure that you can do it before you necessarily ask AI to help you with it.

But this real focus on writing as a core part of thinking: we’ll see how that changes over time. I really appreciate that all of this is subject to a lot of revision, and probably very rapid revision, at the rate we’re going. But yeah, definitely a sentiment to hold onto, I think.

Dan Baer: One of the things I most love about the evolving conversation around AI is the way that this set of new technologies forces us to reckon with some very old philosophical questions in new ways.

As I think about the role that writing plays in my own life, or has played in my own life, I say in the piece that writing has two products. One is the output and the other is the expert. In my experience, it’s the way that you become a deep thinker on a particular topic.

Writing is, for me, an act of thinking, a manifestation of thinking. In that sense, it’s very essentially, in a Cartesian sense, very essentially human. I write, therefore I am, is only a transliteration. And I take that seriously.

I think that for those of us who have been fortunate enough to make a profession out of thinking and researching—maintaining the core of that skill set and that practice—identity is impossible to imagine without the act of writing.

Kristin Hayes: Yeah. And I’m going to show my nerdiness here for a second, but writing can also be extraordinarily joyful. I wouldn’t want to give that up. When everything kind of comes together in that sense of capturing well what it is that you want to say—but also working with others to refine the meaning of things—there’s a lot of joy that comes from the act of writing, a lot of depth and meaning. And it is very human.

Dan Baer: Absolutely.

Kristin Hayes: I think we should hold on to that.

Dan Baer: In addition to having the two outputs I highlighted, it has both a utilitarian and an aesthetic value.

Kristin Hayes: For sure. Well, let me ask you about my very favorite part of the piece, which is where you talk about convenings and the growing value of people who have a knack for bringing communities together for meaningful conversations.

I think that’s probably a part of many of the organizations’ missions that we’re talking about; certainly for Carnegie, for RFF, and I imagine a lot of others. It is, of course, about the research and the writing, but a lot of it is about convenings and how we bring people together.

Full disclosure: this part of the RFF portfolio is what I spend a lot of my time thinking about. So, it’s a little bit of a self-interested question, but you argue that that convening role is going to matter more in an AI world. I’d love to just ask you why, and how you think it will, or maybe it even should, change the types of events that we and our organizations focus on.

Dan Baer: Convenings predate the written word as a way of transmitting information, obviously. So, this is an “everything old is new again” intuition.

I think at the core, my hypothesis is that we will find that … Many of us already do find that human interactions are the kinds of interactions that are often most generative at certain points in the process of thinking through a hard problem, that it is actually easier to think with friends than it is to think on your own. And that doesn’t mean that solitary research and writing won’t remain part of the process, but that actually, there’s a special value to being able to be in dialogue with a group of others. Part of the evidence for this is the way that most of us who use LLMs use them as an ersatz version of dialogue in order to generate outputs.

And I guess my thinking here is that think tanks, in their kind of stereotypical form—at least if you think of the latter half of the twentieth century, a think tank convening was a place to transmit information or research created by the think tank, principally. There has been an evolution of formats; most deplorably, in my view, the panel discussion, where everybody knows exactly what to expect. It’s going to be four people who try to compress months or years of their research into a six-minute opening statement that always runs three minutes too long and is a little bit too complicated for somebody to understand just listening for the first time, and doesn’t necessarily build on what the person before them just said. And then there will be a bunch of comments dressed up as questions from the audience, and then we will eat a stale pastry and go home.

And I think that if you start to think more purposefully about using convenings as a way to generate ideas—not just to transmit them, which again, this is not new. This has been true throughout human history. But if you start thinking more purposefully about how we in the knowledge economy can do that, it forces a reconsideration and an intentionality about curating who the participants are at a particular convening, about breaking away from the panel discussion and thinking about new formats of interaction that really do challenge people to problem-solve together or to iterate on things that don’t require them to necessarily be an expert on a topic in order to be participating at a convening on the topic, because maybe they bring a fresh perspective from another field.

So, I think this is a moment of opportunity, and I think those experiences will increasingly feel special and valuable to people, and therefore be fulfilling and satisfying to people who are in this space. I think the organizations that are able to develop approaches to convening will have something to offer the world beyond the outputs of their research that will be a real value add.

Kristin Hayes: Yeah, and I guess I’m wondering … Most of the folks here at RFF, for example, are PhD economists. They were deeply, deeply trained in economics in ways that continue to blow my mind. I can say this as not one of them! I’m amazed at the depth of economics expertise that exists within my colleagues—but they weren’t trained to host convenings. It’s a different skill set.

So, do you think of it as a skill set that’s important to empower the same researchers who are doing that deep research with? Is this a parallel track? Is this some combination thereof? How does this tie back into our skills-building conversation from before?

Dan Baer: I think it’s TBD, and I could imagine it playing out in different ways.

I think convening can definitely be a team sport and you can have … Certainly in my own life, there are a couple of my colleagues who I think are just really good at thinking through how it’s going to feel in the room once we put this question out to people, and how are they likely to respond? When are they going to need a coffee break? That kind of thing. Who could add a voice that would actually challenge conventional wisdom on this without turning this into an “overturning the table” exercise?

And I will lean on them. I’ll pick up the phone and say, “I’m trying to put together a workshop on this topic. Can I talk it through with you and get your feedback?” So I can definitely see it being a team thing.

I also do think one of the things I said in the piece is just that the kind of successful monastic researcher, who stays in their office and writes a brilliant paper once every year or six months and publishes it and lets it go out into the world without much attention and moves on to the next piece of research, is just going to become more and more anachronistic.

And increasingly, I think researchers—the ones who make the most splash in the world (and that’s not necessarily required for impact, but increasingly, the credentialing effect of making a splash in the world does connect to impact), will be people who are what I call “charismatic intellectuals,” who are able to make ideas “pop,” who are able to be charismatic messengers for ideas.

Kristin Hayes: Interesting.

I’m going to take a pause for a second and just let our listeners know that I’ve only gotten through two of Dan’s intuitions because I’m having too much fun having this conversation. So, forgive me if I run a little bit long, dear listeners, in this episode.

But let me actually take this moment to then move on to the third intuition, which reads, “In our post-expert world, experts may yet come back into favor.”

I feel like we talked a little bit about this at the outset, and I will say, for the record: I very much share this intuition. I think you characterized this much more eloquently than I’m about to, but it has always struck me that, as the volume of information becomes greater and greater, and its quality arguably less and less certain (we can debate that), there still will be a role. People will perhaps need to lean on trusted humans—maybe even more, rather than less.

Let’s just talk a little bit more about this aspect, your theory here, what that might look like in practice. Are we going to be fact-checking those AI-generated explainers, or are we going to, as humans, have something more personal, more interesting—what’s our role going to be?

Dan Baer: I think we very well might be fact-checking the outputs of LLMs, if not literally, then metaphorically.

I think there will be a demand signal when people are swimming through surplus, without expertise themselves, [searching] for a trusted source as a guide, as a Sherpa through that surplus.

The way I’ve been thinking about it is that, increasingly, the successful expert or scholar will be responsible for mediating the intersection between knowledge or information and structures of power—whether those are in government, business, media. There’s going to be this sweet spot, where people in positions of power need to have good information in order to make decisions. And I think there will be a role, for those who can be credentialed experts in a particular field, to help them sort through information that they may already have access to because of the proliferation of content that’s available, but again, may not know how to navigate. And some people will be content to swim in the AI slop, and others will want a life raft. I think there’s an opening for human experts in that world.

Kristin Hayes: I certainly agree and hope so.

Your fourth intuition speaks to how AI might allow those researchers who are primarily, maybe, qualitative researchers to pull more quantitative analysis into their work.

So, you already mentioned earlier: You’ve been pulling in data and creating new data sets. I think we already do a lot of quantitative work here at RFF. I can see this maybe being less of a change for us, but maybe not. I don’t know. And I’d love to ask you for an example of how you see this playing out at a policy research organization like Carnegie.

Dan Baer: There are a bunch of different directions that one can take this point.

At the highest level, AI tools already allow, say, an individual researcher to do complex and large data analysis in a shorter amount of time than they could have. (And I expect them to increase that ability.) They just change the economics of bringing large data sets to bear on what we would broadly call “political science” or social science research. And that is something that has been happening for several years; it’s continuing; et cetera.

I think the follow-on from that, or one of the areas that I’m interested to see how it evolves, is that in many ways in an AI economy (and this is certainly true in the private sector), the moat is decreasingly associated with technology or having access to technology that can do things, and increasingly associated with data and the data that you have, the quality of the data you have, the exclusivity of the data you have.

For example, being able to marry some original data collection with publicly existing data sets will create new data sets that could be a real, defining, unique source of insight into different ways the world works—whether that’s for a PhD economist at RFF, or a PhD political scientist, or a Russia expert here.

I think one of the areas that I’m interested to see is how what used to be mainly argued through more anthropological field work and qualitative research—can we marry some of that with quantitative data that is available, and basically produce a both qualitative and quantitative set of backup four core arguments about how the world works or is likely to work?

The broader point here is that, as good writers (and I speak for myself as well as others, I think), we spend a lot of time being worried about how AI is going to pollute the craft of writing. I make a comparison to fashion houses at the advent of the sewing machine who would’ve said, “No, we can’t possibly use sewing machines. That ruins the handmade nature of our perfect craft.” And it’s important to also think about, What do the sewing machines allow us to do that we couldn’t have done before? What new shapes and materials can we use that we couldn’t have used before? And I think as researchers, we should think about, What do these tools allow us to do with respect to our research that we couldn’t have done before, either for reasons of cost or capability?

Kristin Hayes: As someone who loves her sewing machine and really weirdly loves hand-stitching things, those sewing machine stitches are a lot tidier. They really are a lot better than my hand-stitching. So, I’m not sure anybody would want us to go entirely back to those days.

As much as I hate to rush through this, because I’d love to ask you more about data—I do want to make sure that we get to the fifth intuition, which you acknowledge might be the most controversial. I don’t want to give that short shrift. And that’s about potential shifts in research transparency.

I’m going to summarize this by saying it sounds like Carnegie (just like RFF) has really been pretty strongly committed to putting its research in the public domain rather than producing content for, let’s say, a single client or behind closed doors. There’s this real drive to have our information go out into the public domain for the public good. But you have a sense that it might be worth revisiting that.

Those are my words, not yours. But let me just ask you: Talk me through that last intuition and if there is a reason to revisit that public-domain nature of the research. Tell me more about that.

Dan Baer: As you said, this is in some ways knowingly, if not intentionally, provocative.

I think it’s important to be clear: I don’t think it’s worth reconsidering our public mission. We exist to help people in positions of power make decisions that are better for the world, better for people. And both of our organizations, and many others, have historically made a commitment to make our research public because we believe that that is the most effective way to achieve that public mission.

I guess I’m not making any recommendation, but more forecasting that I could imagine a world in which there’s increasing pressure to ask whether it indeed does advance our public mission to make all of our research public. There are a number of ways that you could explore that.

One is if, say, I have an essential paragraph that summarizes a core idea that could make the United States more secure with regard to its posture in Asia over the next 20 years. If that paragraph is much more likely to be digested through LLMs, and spit out in sentence fragments in various places that are not contextualized the way that I would intend them delicately to be contextualized, I might be concerned that that’s like giving a quote to a reporter where they take the first four words that you said, but not the last four, and the last four change the meaning.

So, you might believe that actually we need to control some of the distribution of our ideas, because we want them to be consumed properly in the way that we intended. And in a world of LLMs, we have less and less ability to predict that the vast majority of people who would be able to consume our ideas will be consuming them in the proper nuance and context that we intended. So that’s one example.

Another is not necessarily that we would hold back any of our core research from being public. As an avenue to impact, going back to the expertise conversation, we may find that the public research is really only a calling card, as it were—it’s not actually informing decisionmakers on its own. It actually needs the human accompaniment of a closed-door meeting or of a series of conversations and conferences that are workshops or whatever in order to make it fly through the world.

So, we may find that more of the impactful work that we do is not fully public. And I don’t mean confidential or classified. I don’t mean that we become spy organizations, but just that more of the ways that we make sure that the ideas land with the audiences they’re intended to, or the audiences that can use them to inform better public decisionmaking—that we rely less on the historical “Here’s a paper that we want the New York Times to write a piece on and that we know that every Senate judiciary staffer is going to read,” and more, “We’re putting a paper out there, and then we’re going to have a bunch of follow-on conversations, some of them closed-door conversations, that really make sure that it lands with the right audience.”

I think there’s a number of different ways you could interpret it. But I do think that some of the information that we have will be valuable, and we need to think about how we want to put it out there into the world. If we synthesize data sets with qualitative research, we may use that to inform our work, but we might make less of it public than we might have at one time. And I think we’ll have increasing questions raised about, How do we make sure that the ideas that we think can really change the world land where they’re supposed to, and how they’re supposed to?

Kristin Hayes: Yeah. Super fascinating. Dan, I will say, periodically I invite guests to come back on the podcast six months later for an update on something. I’m definitely going to extend that offer to you. I feel like this is absolutely a conversation that I want to have maybe every six months, I don’t know, and just test how all of this is unfolding, because we’ve both acknowledged it’s moving so fast—the whole world seems to be moving so fast; certainly the space of AI is.

But I think one of the reasons I was so drawn to the piece is because these feel like really important intuitions for places like RFF, things that we can be keeping an eye out for, and just kind of buoys in the stormy seas a little bit. I really appreciate your time talking through them. You have an open invite to come back, to do some evaluation of how things are looking.

Dan Baer: I hope I’ll have something different to say! As you said, it would be a real failure if nothing’s changed. That wouldn’t reflect the state of the technology—it would reflect the state of my thinking.

Kristin Hayes: You can come back and gloat if you just nailed them all, but I imagine there’s going to be a lot more to learn.

Well, Dan, let me close with our regular feature, Top of the Stack. Would love to ask you to recommend some more good content. Could be on this subject, could be completely random, but anything you’d want to recommend to our listeners. So Dan, what’s on the top of your stack?

Dan Baer: I will recommend both a nerdy thing and an artistic thing.

I’m reading a dual biography right now called The Wonderful World That Almost Was, which is about two mid-twentieth-century artists, Peter Hujar and Paul Thek, a photographer and a painter-sculptor. And one of the things that I’m loving about that is kind of an image of, in the art world, two artists among a community of artists who are challenging prevailing wisdom; and classical ideas; and what it felt like, two generations ago, to do that. And I’m enjoying thinking about what the equivalent is today.

And then there’s a piece that I’ve been reading over the last week that is a new economics article called “Terra Incognita: The Economics of a Shrinking World,” which is by two professors, one from Penn and one from Northwestern, that makes the point that many countries have been talking about demographic decline and the fact that their populations are not growing. But as a planet, we are reaching a threshold where the human population for the first time in a long time (aside from catastrophic events) will not be growing. And we’ve never really dealt with that before. And it has all kinds of knock-ons for the political economy of a potentially shrinking population on the planet. And we are approaching that point, and these authors try to tease out why that is and what it might mean for the way that our world is shaped. And I think those kinds of big political economy questions are really interesting for thinking about the future.

Kristin Hayes: A lot of stuff to think about for the future, for sure. I love these conversations. Dan, thank you so much. It’s been a real treat. I appreciate it.

Dan Baer: Thanks so much for having me.

Kristin Hayes: You’ve been listening to Resources Radio, a podcast from Resources for the Future, or RFF. If you have a minute, we’d really appreciate you leaving us a rating or a comment on your podcast platform of choice. Also, feel free to send us your suggestions for future episodes.

This podcast is made possible with the generous financial support of our listeners. You can help us continue producing these kinds of discussions on the topics that you care about by making a donation to Resources for the Future online at rff.org/donate.

RFF is an independent, nonprofit research institution in Washington, DC. Our mission is to improve environmental, energy, and natural resource decisions through impartial economic research and policy engagement. The views expressed on this podcast are solely those of the podcast guests and may differ from those of RFF experts, its officers, or its directors. RFF does not take positions on specific legislative proposals.

Resources Radio is produced by Elizabeth Wason, with music by RFF’s own Daniel Raimi. Join us next week for another episode.