This study aims to map institutional responses to AI adoption in universities worldwide by analyzing a large text-based data source: university websites. As a key communication tool, websites reflect an institution’s culture and organizational norms (Fowle & Vassaux, 2017), shaping how universities engage with internal and external audiences.

We employ web scraping techniques to collect AI-related content from the websites of 2,000 universities included in THE rankings. Our analysis focuses on the scope and thematic dimensions of AI-related discussions in university communications. In the first stage, we apply text mining techniques, including keyword extraction and topic modeling using NLP approaches. We then classify and describe university profiles based on AI engagement, incorporating institutional characteristics from databases such as ETER.

Our approach driven by exploratory inquiry, we attempt to address the following questions:

• What types of universities fall into the "wait-and-see" category in response to rapidly advancing AI technologies?

• To what extent do universities address AI-related topics beyond ethics and teaching and learning? What is the mix of topics that appear on the websites? What typical university profiles do they form?

• Which stakeholders do universities prioritize in their AI-related communications through their websites - prospective students, labor market partners, the knowledge and innovation industry, or policymakers?

The advent of widely available technologies such as electricity, the internet and search engines, has already reshaped higher education many times. For this study we focus on three assumptions of institutional context related to AI technology in higher education that have become widely accessible since 2023. 

Firstly, there is the rapid adoption of genAI technologies by users. In the words of technology acceptance models (David, 1987; Teo and Noyes 2011) perceived ease of use, perceived usefulness and perceived enjoyment might positively contribute to genAI technologies usage among students (some empirical evidence Cano and Nanez 2024). This encourages an 'organic' spread across the higher education sector (Frumin, 2024), which institutions must deal with. 

Secondly, the market for genAI-integrated applications is thriving -  the development of numerous general-purpose tools (like ChatGPT)  as well as specialized applications (e.g. Copilot); also targeting specific needs in higher education, from teaching and learning tools (like Cramly) to university service management (such as Ivy.ai). Universities are also stimulating the development of new AI models. Although the variety of tools provide a choice, it also forms the chaos of on-going changes.

Thirdly, universities are navigating the complex dynamics between multiple actors - students, faculty, university management and national policy. For national policies AI becomes a strategic issue that calls for varied intervention approaches across different countries, often framed as a “new space race” (Ulnicane et al 2021). Universities are not only shaped by public policy - through regulations, funding mandates, and accountability measures - but also act as active contributors to policy development, especially among elite institutions.

Previous research on universities’ responses to AI highlights key takeaways that provide a foundation for our large-scale data exploration. Korsberg and Elken (2024) show, based on interviews with university leaders, that many institutions remain in a state of hesitation, reluctant to adapt despite pressure from the media, students, and emerging technological trends. Then, recent research on AI-related university policies across the Global North highlights three dominant themes: guidelines for ethical generative AI use, the design of assessments, and training programs for faculty and students to enhance AI literacy (Jin et al., 2025). Moreover, at the institutional level, universities prioritize responding to labor market demands and employer expectations. Analytical reports indicate a global increase in the availability of AI-related educational programs and the number of their graduates (Maslej et al., 2024).

To analyze AI-related content from the websites of 2,000 universities included in Times Higher Education (THE) rankings, we utilize a combination of web scraping and text mining. Our analysis aims to uncover the scope and thematic dimensions of AI-related discussions across a wide range of university communications through websites. The process is divided into several stages. In the first stage, we employ web scraping techniques to collect data from the universities' websites. This involves using Python-based tools such as Requests and Beautiful Soup t. Selenium is used to automate interactions with dynamic websites that require JavaScript execution, ensuring that we capture all available content. Once the data is collected, we move on to text mining. We focus on two main techniques: keyword extraction and topic modeling. For keyword extraction, we use algorithms such as RAKE (Rapid Automatic Keyword Extraction) and KeyBERT. RAKE is effective in identifying key phrases based on word frequency and co-occurrence patterns. We incorporate KeyBERT, a more advanced method that utilizes contextual embeddings to provide more accurate and semantically meaningful key phrases. The next step involves topic modeling, where we apply NLP approaches. By identifying the keywords and topics of discussion, we can map the thematic landscape of AI-related content across universities and analyze how these themes vary according to geographical location, institutional profile, and other factors. We proceed to classify and describe university profiles based on their AI engagement. We use clustering techniques to categorize universities into different engagement groups, considering factors such as the frequency and types of AI-related content. Additionally, we integrate institutional characteristics from external databases such as the European Tertiary Education Register (ETER).

We expect that our analysis of AI-related content on the websites of 2,000 universities will reveal key themes such as AI research, ethical guidelines for generative AI, assessment design, and AI literacy programs for faculty and students. This builds on previous research by Korsberg and Elken (2024), who found that many universities are hesitant to adapt to AI, despite growing pressure from media, students, and technological advancements. Our analysis will help identify how universities are framing AI adoption, either as a tool for innovation or as a response to labor market demands, as highlighted by recent studies (Jin et al., 2025). Given that our sample consists of universities striving for world-class status, we expect that their engagement with AI will reflect a variety of institutional priorities, including addressing societal challenges, positioning themselves as leaders in technological innovation, or collaborating with industry. We anticipate uncovering patterns in AI engagement that highlight differences between regions and types of universities, with some focusing on AI research while others prioritize workforce development through AI-related educational programs.

Jin, Y., Yan, L., Echeverria, V., Gašević, D., & Martinez-Maldonado, R. (2025). Generative AI in higher education: A global perspective of institutional adoption policies and guidelines. Computers and Education: Artificial Intelligence, 8, https://doi.org/10.1016/j.caeai.2024.100348 Maslej, N., Fattorini, L., Raymond Perrault, Vanessa Parli, Anka Reuel, Erik Brynjolfsson, John Etchemendy, Katrina Ligett, Terah Lyons, James Manyika, Juan Carlos Niebles, Yoav Shoham, Russell Wald, & Jack Clark. (2024). Artificial Intelligence Index Report 2024. AI Index Steering Committee, Institute for Human-Centered AI, Stanford University. https://aiindex.stanford.edu/wp-content/uploads/2024/05/HAI_AI-Index-Report-2024.pdf Davis, F. (1987). User acceptance of information systems: The technology acceptance model (TAM). Available online at: https://umich.edu/ Korseberg, L., Elken, M. Waiting for the revolution: how higher education institutions initially responded to ChatGPT. High Educ (2024). https://doi.org/10.1007/s10734-024-01256-4 Ulnicane I, Knight W, Leach T, Stahl BC, Wanjiku WG (2021) Framing governance for a contested emerging technology: insights from AI policy. Policy Soc 40(2):158–177. https://doi.org/10.1080/14494035.2020.1855800 Teo, T., & Noyes, J. (2011). An assessment of the influence of perceived enjoyment and attitude on the intention to use technology among pre-service teachers: A structural equation modeling approach. Computers & education, 57(2), 1645-1653. https://doi.org/10.1016/j.compedu.2011.03.002 Frumin, I. (2024, July 24). “Managed” and “Organic” Diffusion of Generative Artificial Intelligence in Higher Education: in Search of a Balance. [Virtual Conference Presentation]. 22nd Annual MEITAL National Conference, Tel Aviv, Israel. https://meitalconf-iucc-ac-il.translate.goog/he/meital2024/international-track/?_x_tr_sl=auto&_x_tr_tl=en&_x_tr_hl=en&_x_tr_pto=wapp Frumin, I., Kiryushina, M., Vorochkov, A., Platonova, D., Terentev, E. (forthcoming) Mapping the Generative AI Research in Higher Education: 2022-2024 Insights Fowle, M., & Vassaux, C. (2018). How entrepreneurial universities walk the talk or win the prizes: an online content and factor analysis. International Journal of Innovation and Regional Development, 8(4), 286-305.