人工智能未来对研究型图书馆的战略影响：研讨会报告

Karen Estlund, Dean of Libraries at Colorado State UniversityCynthia Hudson Vitale, Associate Dean at Johns Hopkins University This work is licensed under a Creative Commons Attribution 4.0 International License. Cover image by Cash Macanaya on Unsplash Table of Contents Executive Summary41Introduction & Context6Introduction to the AI Scenarios & Strategic Planning7How the Scenarios Were Developed7Workshop Setup8Starting Assumptions9Scenarios Overview12Finding Direction Among Uncertainty: Robust Strategies14Robust Strategies15Five Bold Steps: Ignite Strategic Momentum Across the Organization22Synthesis: What the Bold Steps Tell Us26Reflections A Year Later27Drivers January 2025–February 202628Signals January 2025–February 202629Critical Uncertainties a Year Later3Strategic Actions a Year Later32Conclusion: Planning Ahead in Uncertain Times42Appendix43Current Library Capabilities in Scenarios43Scenario-Specific Strategies45Strategic Implications Workshop, Futurescape Libraries: Mapping Possibilities forTomorrow’s Information Hubs56Attendee List58AI Tools Used in the Construction of this Report60 Executive Summary Research libraries are navigating one of the most consequential periods oftechnological change in their history. Generative AI is no longer a discrete toolcategory; it is embedded in discovery systems, publishing platforms, researchworkflows, and enterprise software, reshaping the economics of knowledge access, the To support strategic planning under these conditions, the Association of ResearchLibraries (ARL) convened the Strategic Implications Workshop, Futurescape Libraries:Mapping Possibilities for Tomorrow’s Information Hubs, on December 7–8, 2024, inWashington, DC. Facilitated by Keith Webster, Helen and Henry Posner, Jr. Dean of Participants worked across four divergent 2035 scenarios to surface strategicvulnerabilities, identify near-term decision points, and develop an agenda that remains •Democratized and Socially Integrated AI•Consumer-Oriented AI•Laissez-Faire AI•Autonomous AI The resulting robust strategies, valid across all four scenarios, are organized into sixthematic areas: Workforce Development and Organizational Culture; Collections,Technology and Infrastructure; AI Literacy and Critical Skills; Ethics and Values-Based Across all scenarios, several near-term priorities emerged with particular urgency for •Invest in workforce now.AI literacy programs, flexible position descriptions,and safe opportunities for staff experimentation are foundational regardless of •Leverage unique collections.Research libraries hold assets—special collections,digitized archives, curated corpora—that commercial AI systems cannotreplicate. Using these to responsibly train or inform local AI models and •Lead on governance and ethics.Establishing AI governance frameworks, ethicsboards, and evaluation processes for bias is not only an institutionalresponsibility, it is a competitive differentiator. Libraries are positioned to lead •Build and deepen partnerships.Forging cross-campus collaborations with AIresearchers, integrating libraries into institutional AI strategies, andstrengthening collective action across the library community on copyright, Since the December 2024 workshop, the environment has continued to shift quickly.Generative AI is now embedded across everyday productivity tools and increasinglyinside library vendor platforms, while early agentic workflow automation is moving AIfrom “assistance” to “action,” raising immediate governance requirements forauthorization, auditability, containment, and quality control. At the same time, de- The two original axes Societal Adaptation to AI and Intentionality in AI Process andDesign continue to hold up as the most discriminating uncertainties for divergentfutures, even as the uncertainty has become more operational and sharply contested inpractice. In the current context, signals increasingly suggest drift toward less A year later, libraries have made the most consistent progress in building internalcapacity by expanding staff training and professional development and beginning to around positioning distinctive collections for AI-era use and securing footholds incampus governance and guidance efforts. At the same time, several critical elementsremain underdeveloped: translating principles into operational controls for vendor-embedded AI, implementing provenance and IP compliance in everyday workflows, These findings confirm that the window for proactive positioning is open butnarrowing. Research libraries that treat AI governance, workforce development, andcollection strategy as integrated priorities will be best positioned to lead their Introduction & Context The Artificial Intelligence (AI) Strategic Implications Workshop (December 2024)used the ARL / CNI 2035 Scenarios: AI-Influenced Futures in the ResearchEnvironment as a structured basis for examining how artificial intelligence andmachine learning could reshape the research,