What is the role of AI in think tanks; and the role of think tanks in the age of AI?

I had the distinct pleasure of attending a #wonkcomms brunch this morning where PTKO CEO Tony Kopetchny and Emerson Brooking, a Senior Fellow at the Atlantic Council, shared an altogether optimistic appraisal of the manyfold and often unprecedented capabilities that AI can bring to Think Tanks. There is no getting around the risks, especially those made poignant by ChatGPT: the risk of misrepresentation, misappropriation, hallucination, etc… but none of these preclude beneficial uses of AI.

Emerson pointed out that AI is nowhere near as new as 2023 headlines would suggest, and in fact, ChatGPT itself emerged as a fairly modest innovation on top of previously developed models. There were rumors that OpenAI decided to develop ChatGPT as a somewhat snap decision and a hail-mary experiment, launching the product just 2 weeks later with no idea that they were about to upend the global perspective on the practicality of AI in the workplace.

My takeaway from that? We are all 2 weeks away from similarly iterating capabilities that we already have access to into a limitless number of innovations. Seizing this opportunity will require experimentation, intentionality, leadership, and certainly caution and corresponding development of governance… There is SO MUCH more to say about this, and it is becoming a big focus of our consulting work at PTKO.

But what of the central question: What has AI done to the traditional role of think tanks?
A few senior think tank communicators spoke to the elephant in the room: media and journalism are already taking the tech industry to court over its use of their content to train the AI. If ChatGPT can output your organization’s talking points in concise and compelling detail without even a nod of attribution, should think tanks similarly be pushing back as well? Particularly when attribution and citation are often tied to the funding that is the lifeblood of these organizations?

I think there is a lot that AI providers can do to be better stewards and more honorable participants in the knowledge ecosystem (Google is already taking steps in this direction), but my adversarial perspective on this is that we can’t put Pandora back in the box. Think tanks need to lean into an AI-driven future because:

  1. Fighting the inclusion of our ideas in AI platforms goes against our mission to make our ideas accessible and influential. AI is going to be a part of how humanity consumes information moving forward, so allowing our content to be used for training will become at least as important as it has been to be indexed by search engines.
  2. Funders of think tanks will need to modernize their metrics, prioritizing things like social listening and legislative outcomes, recognizing AI channels as a broken link in the chain of custody of knowledge that makes legacy metrics inexcusably imprecise. Saying this is no help to a think tanker who is dependent on a recalcitrant foundation, but highlights that there are alternatives we can offer if our traditional metrics lag.

Think tanks do not have a monopoly on knowledge. This has always been the case vis-a-vis traditional media and later social media, where our goal is to see our ideas take root and flourish, even as we hope to earn a citation. Recognizing AI products as a new marketplace for knowledge, our goal should be to influence them just like we do any other channel.

So what becomes of the role of think tanks? I think it comes down to 3 things…


Think tanks do not have a monopoly on knowledge, but they do hold a reputation as an incredible asset: Information generated by an AI will be received with a healthy dose of skepticism. That same information will be trusted as fact if validated or shared by a think tank with its core audiences. Tony addressed this with his idea of a Think Tank consortium that could manage a shared watermark of authenticity for ideas produced by think tanks. Imagine a similarly managed AI plugin that could automatically assess and validate AI-produced “facts” as they are read off by Alexa, extending their reputation well beyond their traditional sphere of influence.


Researchers and communicators have watched with fear as ChatGPT spits out a position piece in seconds that might have taken staff hours of labor and perhaps days or weeks of available calendar time to produce. And they then may watch with horror as the same ChatGPT writes an equally compelling piece opposing those exact same views. These models are making content cheap. What they cannot replace is the purpose that we bring to our work. That underlying intent and vision of the world that our think tanks are advocating for is the true purpose of a think tank and its staff, and the looming war of AI-generated words only deepens the importance of our humanity as the principled core of these organizations.


As for Emerson’s answer to the question of if think tank strategy should respond to AI? Let’s get back to our roots. Think tanks evolved as spaces for literal meetings of the minds, providing venues for academics and policymakers to form trusted relationships, enlighten one another, and call upon one another for urgent responses to complex challenges. This strength and impact didn’t come from knowledge itself, but the ability to transfer it, often person-to-person. There’s no such thing as a free lunch, except at a DC think tank, and during the pandemic think tanks were forced to innovate and move their relationship-building practices into new media. Continuing to invest in and diversify this strength will secure think tanks’ future in an influence landscape overrun by AI.

Download the slides from this presentation.