Ashley Barey (VP Product Management, Clarivate) · 4:30–5:30 PM · Denver Room · General Track · Tuesday, April 14
Speaker: Ashley Barey
Ashley Barey presented Clarivate’s approach to AI across the Innovative product suite — the Responsible AI framework (Transparent, Ethical, Safe), current and upcoming AI capabilities, and data from the 2025 Pulse of the Library survey.
Ashley opened by challenging the common comparison of AI to cloud computing or Google search, arguing that AI is fundamentally different in three ways:
A bad Google search just returns wrong results — you can tell. A bad AI output can come across as very convincing and fluent. Not knowing whether the AI is wrong is the real danger.
AI reasons, makes decisions, and builds on prior context — it’s not just infrastructure and retrieval. Google search doesn’t assume anything; AI is asking and answering questions.
AI raises all the prior concerns (antitrust, privacy) plus new ones: bias, autonomy, misinformation at scale, and existential risks.
Ashley referenced jobs and AI displacement — reskilling and job transformation, not wholesale replacement. 📎
Ashley showed a slide of book covers and made the point that our collective perception of AI is deeply shaped by fiction — from Samuel Butler’s Erewhon, to Asimov’s Three Laws of Robotics, to Arthur C. Clarke’s HAL 9000, to The Terminator. 📎
“Your patrons have been watching all of that too. They don’t know how to interpret AI. Libraries are the best civic hub — the place to go to get good information, to use those trusted advisors for teaching and leadership.”
Data from the Clarivate Pulse of the Library 2025 survey (2,000+ librarians across 109 countries; 400+ public library responses). 📎
“Where does the information go?” — the top concern across public libraries. Use of public LLMs with library data is a governance issue.
Came up at PLA from 3 different customers, largely in areas where data centers are being built. Ashley offered a counterpoint: data centers were already being built during the Big Data era, and AI may actually help reduce environmental impact through efficiencies (e.g., supply chain emissions reduction via prescriptive analytics).
All current Clarivate AI features (Data Explorer, Metadata Assistant, etc.) are included under existing licenses — no extra cost. Ashley acknowledged this is a “golden age” of cost vs. value.
Key quote from survey: “Getting beyond initial exploration and into problem solving with AI will therefore be essential to libraries taking a positive long term strategic approach.”
Libraries that have identified someone to champion AI — making decisions, setting guidelines, working on policy — are seeing more successful adoption.
Clarivate has invested heavily in internal AI training over the past year. Libraries seeing success are doing the same — upskilling staff to understand business problems and apply tools effectively.
Don’t dive in with scattershot AI adoption. Understand your business problems first, then run focused proof of concepts. It’s OK if a POC isn’t successful — that’s the point.
Clear indication of AI features. Clear information about what data is used and how. Data is not stored with AI agents.
AI with a purpose — solving real problems, not playing around. Measures to reduce bad information. Collaboration with industry organizations and the Customer Advisory Board on responsible AI implementations.
Human in the loop. Uphold privacy & security standards. Adherence to evolving global regulations. Referenced the NIST AI Risk Management Framework. 📎
Launching in the latter half of 2026. Clarivate’s academic side already has an AI Advisory Board; the public board follows that model.
AI-assisted content generation for newsletters and community outreach. On-demand AI image generation.
AI-enhanced web experiences (available as an option).
Generate SQL via natural language search. Early access coming by end of year.
Generates MARC suggestions, saving libraries 20–180 minutes per record.
Conversational AI for reporting and analytics. Supports AI for Polaris and Sierra as well.
Showcase generation. Natural Language Search and Chat POCs.
Handles purchasing and invoicing workflows.
Clarivate’s academic side is ahead on AI adoption, and the public side benefits from their learnings. Key product:
Nexus — a browser extension that works inside ChatGPT, Claude, and Gemini. Scans AI responses for scholarly references and verifies them against Web of Science, ProQuest, and Primo/Summon. 📎
Audience member commented on the value of achieving a failure state quickly — it’s much quicker to develop and test ideas with AI. “What didn’t work is sometimes a lot more valuable than anything else.” Ashley agreed and noted AI is also excellent for problem refinement: “Here is the problem I’m having — what are the data points or KPIs I need?”
Question raised about a “bring your own agent” model — making catalogs more agent-friendly rather than having agents scrape them in uncontrolled ways. Ashley acknowledged this is a security concern: “These things are so hungry for data, and your catalog is an obvious source.” Need guardrails to prevent agents from bringing down existing infrastructure.
Will AI features eventually cost extra? Ashley: “We’re in a golden age of cost vs. value.” AI providers are already “turning up the dial.” Current features won’t have added costs, but more advanced capabilities may. “We’re very hesitant in the environment we find ourselves in.”
