Scenario Planning for the AI-Driven Future of Research Libraries

It’s no surprise that artificial intelligence dominated the annual Center for Networked Information (CNI) meeting this past December in Washington DC. CNI’s long-serving executive director, Clifford Lynch, has done as much as anyone in the US research library community to celebrate, embrace, and inform others about AI. While grumblings of AI fatigue could be heard in the CNI conference halls, AI research projects are winning Nobel Prizes, and generative AI tools continue to penetrate and improve academia; those in attendance had little choice but to sit up and pay attention.

Libraries, especially research libraries, must confront AI and its impact on the ecosystem of knowledge, research, and learning. But it remains unclear how best to plan for AI when its opportunities and challenges are shifting so rapidly. AI touches all facets of a research library’s operation, from research to learning to workforce development. Is there a way to future-proof libraries’ role in ensuring high-quality research and student engagement? Can key players keep their eye on what matters most while navigating unpredictable political realities and technological breakthroughs? Is it possible for research libraries to proactively shape rather than just react to what the future brings?

The accelerating impact of AI, as well as a pandemic that caught many organizations flat-footed, has prompted renewed interest in futuring to answer exactly these types of questions. While there are many different futuring approaches, what they all offer are techniques and practices organizations can use to prepare and respond to uncertainty.

Our confidence in predicting the future decreases relative to time. This challenge, which faces every planning effort, helps to explain why strategic planning does not usually extend beyond five years. Futuring approaches, like scenario planning, however, are designed to take growing uncertainty into account, helping organizations to plan ten, even twenty years into the future.

In October of 2024, the Iowa State University Library collaborated with Iowa State Extension and Outreach to design and facilitate AI scenario planning, a process that was relatively new to both organizations. The goal was to help the library confront the difficult questions and opportunities surfacing around AI and take a wider view of the impact it is having on their work.

What exactly is scenario planning? Often institutions are merely reactive to changes. Even when a planning process is implemented, it tends to proceed based on an imagined future full of unstated assumptions. The idea of scenario planning is to be more deliberate in identifying driving forces, uncertainties, and critical variables that could shape various futures—not just in the near term but over the next ten years. Rather than taking for granted a conflicting mix of best-case and worst-case scenarios, participants work through different possible futures that might come to pass. This allows them to identify strengths, opportunities, and strategies for the scenarios, each representing a unique version of the future. They also identify indicators to monitor that will signal which scenario or parts of a scenario are unfolding. Such analysis helps them to see what strategies and insights apply across scenarios and to lay the groundwork for more nimble and informed decision-making. The institution is empowered to be more proactive, capable of not just responding to but shaping a fast-changing environment.

The standard way of beginning a scenario-planning process is for the institution or organization to develop its own scenarios after identifying drivers (technological advances, economic trends, regulatory changes, social and environmental issues) and critical uncertainties (government policy, market behavior, cultural reactions). The ISU Library decided instead to draw on a scenario-planning document recently published by the Association of Research Libraries and CNI, as it was timely and thoughtfully constructed and saved them time-consuming work (Association of Research Libraries, Coalition for Networked Information, and Stratus Inc., 2024).

The ARL/CNI scenario set is conceived along two axes of possible influence: “intentionality in AI process and design” and “societal adaptation to AI.” Will these variables be extensive or limited?

Four scenarios populate the quadrants of the bisecting axes:

• Democratized and Socially Integrated AI (extensive intentionality in AI design and extensive societal adaptation to AI): “a world in which an extraordinary convergence of advances in human-computer interfaces and AI technologies create an unprecedented integration of human and computational capabilities that flourish with increasingly open knowledge access.”
• Consumer-Oriented AI Focused on Education and Entertainment (extensive intentionality in AI design but limited societal adaptation to it): “a world in which AI’s impact on the research and knowledge ecosystem is relatively low with the primary AI advances and impact being seen in consumer applications that are readily profitable, relatively uncontroversial, and lower-barrier applications.”
• Laissez-Faire AI (limited intentionality in AI design and limited societal adaptation to it): “a world of missed opportunities, bad decisions, and fecklessness, punctuated by a somewhat haphazard assortment of commercial or other kinds of successes” in which “the … hype around AI and the belief that AI will be the solution to the world’s most difficult problems results in an overzealous and hasty adoption of AI in both consumer life and professional applications.”
• Autonomous AI (limited intentionality in AI design but extensive societal adaptation to it): “a world in which AI is becoming an increasingly independent partner and collaborator in research and learning, leveraging the expanding open resources and data, and also exploiting the scarcity and high cost of human resources” and where “society … has knowingly and unknowingly given up increasing agency to AI.”

In the ARL/CNI document, each of these scenarios is described in further detail and accompanied by an account of how each future is experienced by a fictional director of a foundation for reviewing and awarding grants to innovative research.

The planning process we designed around the ARL/CNI scenarios included a two-hour online introductory session and two three-hour in-person workshops, each facilitated by Iowa State Extension and Outreach specialists. Approximately 25 participants were selected from across the library to ensure a broad array of backgrounds and expertise. Since most participants were new to this type of planning, the initial online session provided an overview of scenario planning as well as a walk-through of the four ARL/CNI AI scenarios. The two- to three-page scenario descriptions were read aloud and then participants made observations and asked questions. Participants were encouraged to put on hold their own predictions about the future as well as their doubts or confidence about the likelihood of any or all scenarios. However improbable it is that any of the four scenarios will come to pass as such, it is likely that parts of each of these scenarios will become reality. Moreover, the playful exploration of possibilities, even improbable possibilities, can be stimulating and revealing.

The subsequent scenario planning process took place over two days, each involving an intensive three-hour session. After a brief orientation to the process, the participants were given time individually to brainstorm opportunities and threats for one of the scenarios. Their ideas were shared to the large group and noted on flipcharts. Next, participants broke into groups of five and discussed what strategies would be most effective, given the opportunities and threats of that possible future. What actions should be taken now? What plans should be developed? The small groups wrote up a summary of their ideas on flipcharts and shared them with the large group. Finally, for each scenario, the large group debriefed about which strategies were best and identified markers of the driving forces and critical variables that the library should monitor. After all four scenarios had been analyzed, the group took a panoptic look at strategies that were robust across multiple scenarios and began to prioritize next steps.

One of the unexpected results of the process was a renewed commitment to continue and to develop much of what the library had already started: building important partnerships across the university, training staff in AI, developing AI literacy opportunities for students, continuing to preserve cultural memory, insisting on and advocating for the integrity of data and knowledge, maintaining human connection and decision-making in relationship to technology, and continuing to position the library as a community-wide convener of ideas and discussion around AI.

But the group also conceived of some new next steps. In working through the scenarios, they decided:

• to assign roles for monitoring the environment for signals and markers of change and to communicate those changes to the library at large,
• to survey students and the public to find gaps in knowledge that could be filled by the library,
• to focus more attention on helping those who lack basic AI skills,
• to pursue new funding sources for work with AI,
• to explore corporate partnerships for free or reduced-price access to AI tools,
• to develop consistent AI learning objectives and to map clear instructional paths,
• to be more sensitive to environmental concerns around AI,
• to build relationships with the humanities and social sciences for the purpose of promoting and developing human thinking independent of AI,
• and to position and promote the library more aggressively for its quality services with regard to information technology.

The outcomes of the scenario planning process helped confirm and orient the ISU library’s AI direction. Efforts are currently underway to incorporate identified action steps and priorities into the library’s strategic goals and framework.

Beyond priorities and goals, however, a secondary benefit of the process also emerged. While caution and anxiety around the rate of AI progress remains, the sessions revealed a deposit of deep curiosity and interest that was expressed throughout the process in lively discussion and excitement. The AI scenarios were very effective at prompting conversation and engagement, which has delivered a helpful boost of momentum to the library’s AI efforts.

The closing plenary of the December 2024 CNI meeting featured Tony Hey, winner of the Paul Evan Peters Award, who spoke about the AI revolution taking place in science. AI-powered breakthroughs are rippling across all disciplines, from medicine to materials to mathematics. This technology revolution is reshaping knowledge creation, and it is reshaping universities and research libraries. ARL and CNI have done the community a favor by producing ready-to-workshop AI scenarios. Research libraries should take advantage of the scenarios to help chart their path in this rapidly evolving landscape.

References

Association of Research Libraries, Coalition for Networked Information, & Stratus Inc. (2024). ARL/CNI AI Scenarios: AI-Influenced Futures. https://doi.org/10.29242/report.aiscenarios2024