What 311 academic librarians from 31 countries told us about AI
The use of artificial intelligence in higher education is sparking heated debates. Do academic librarians embrace AI tools or are they rather wary of them? The results of our most recent Insights study reveal a mixed picture across the world.
Nearly 90% of academic librarians report that artificial intelligence (AI) is now considered a legitimate tool at their institution – though only 47% say their organisation has developed formal guidelines on its use. Behind that headline figure, the picture is more varied than it first appears.
De Gruyter Brill’s Insights team commissioned research company Gold Leaf to survey 311 academic librarians across 31 countries, supplemented by qualitative interviews with 12 senior librarians worldwide. This is what the data showed.
How librarians are using AI today
Globally, the most common uses of AI among academic librarians are writing and editing support, summarisation and explanation of text, search and discovery (often as a first pass in research), teaching and AI literacy and metadata generation and internal library workflows.
Usage patterns differ significantly by region. In Canada and the UK & Ireland, a substantial proportion of responses indicated hostility or non-engagement – nearly forty percent of UK & Ireland responses were blank or negative. South America, by contrast, produced no hostile responses, with some librarians describing sophisticated uses such as mental mapping and finding relationships between ideas. In Europe, there were more references to translation services and specific library functions. India and the APAC countries showed pragmatic, broad-based uptake, with particularly strong use of AI for writing and search.
The tools librarians use
Microsoft Copilot was the most commonly mentioned AI tool globally (68.4%), followed by ChatGPT (54.6%) and Turnitin (40.3%).
Regional differences were notable. Copilot usage was highest in Canada (93.8%) and the UK & Ireland (87.3%). ChatGPT was most popular in South America (83.3%) and least used in the UK & Ireland (30.9%). DeepSeek had notably higher uptake in India and the APAC countries (34.5%) compared with other regions.
Who’s paying for AI (and who isn’t)
The research shows that most institutions using Copilot are paying for it, reflecting its enterprise licensing model. ChatGPT tells a different story: it is widely used but far less often paid for, consistent with the availability of its free tier. Publisher-developed AI tools are used by around 39% of respondents globally, with payment levels varying significantly by region.
Two thirds of respondents (66.5%) said their institution is not currently considering purchasing additional AI tools.
Where librarians see the most potential
Despite the caution, there is strong interest in specific AI applications. Translation tools drew the highest global support (79.8%), followed by semantic search (74.7%), citation and formatting tools (72.1%), literature reviews (69.1%), AI-recommended search (66.3%) and plain language summaries (65%).
In teaching, nearly seventy per cent saw potential for AI to enhance writing and 55.7% for assisting learning methods more broadly. These findings highlight strong interest in AI applications that support research, discovery and learning.
De Gruyter Brill will continue to engage with the academic library community as AI use in research, teaching, and library services evolves.
[Title image by kali9/iStock/Getty Images Plus]