Kudos has released the Phase I report from its Taming the Crocodile project, showing that artificial intelligence now accounts for a third of scholarly information‑seeking. The study, based on surveys of 11,500 researchers and 200 librarians and an environmental scan of more than 300 sources, highlights the rapid shift toward AI‑mediated discovery and sets out recommended actions for publishers to maintain visibility through new channels.
Key findings include evidence that one in ten researchers begin their searches directly with AI tools, while another 25% start with general search engines where AI‑generated answers are embedded. Early‑career and corporate researchers are twice as likely to rely on AI tools. The report also noted that 40% of researchers mistakenly believe AI‑generated overviews are curated by humans, with librarians reporting frequent requests for articles that do not exist due to AI hallucinations. Declining usage of primary sources was observed, raising concerns about subscription sustainability.
The study recommends that publishers develop structured, plain‑language summary content optimized for AI discovery, implement schema and markup for AI readability, and provide educational materials for librarians and end users. Additional guidance includes strategies for generative engine optimization, AI visibility tracking, and maintaining discoverability in global markets such as China. The report also suggests repositioning publishing value toward curation and stewardship of knowledge, supported by cryptographic trust mechanisms.
Industry sponsors emphasized the importance of acting on these findings. Emilie Delquié of Silverchair noted that the evidence provides a foundation for adapting industry infrastructure. Shane Rydquist of CACTUS Communications observed that researchers rely on AI but do not fully trust it, making evidence and attribution critical for credibility. Dan Penny of Springer Nature stressed the need for rapid and responsible action to safeguard scholarly communication. Other sponsors included Elsevier, Emerald, and the Royal Society of Chemistry.
Phase II of the project will focus on coordinated responses, including benchmarking AI’s impact on usage and revenues, observing researcher behavior, and developing shared standards for content and machine‑readable signals.
Click here to read the original press release.
