Intelligent Textbooks
The Adaptation and Personalization Community Group intends to explore and discuss technologies including those to enable intelligent textbooks.
Textbooks have long had a fundamental limitation as a one-size-fits-all medium. Building upon decades of research into their advancement, using state-of-the-art and emerging artificial-intelligence techniques, textbooks can be greatly enhanced.
In Intelligent Textbooks, Sosnovsky, Brusilovsky, and Lan (2025) indicate that the evolution of intelligent textbooks has unfolded in five generations since the 1990’s. These are: “intelligence engineered”, “intelligence integrated”, “intelligence extracted”, “intelligence data-mined”, and “intelligence generated”.
With respect to the first generation, “intelligence engineered”, intelligent textbooks were born at the crossroads of intelligent tutoring systems and adaptive educational hypermedia. After the advent of the Web, intelligent textbooks explored a wide range of personalization techniques to support readers: navigation support, adaptive page manipulation, content recommendation, content sequencing, and combinations of different techniques.
The second generation, “intelligence integrated”, explored open architectures, integrations with external resources, and standardizations of semantic models for supporting these scenarios. This generation sought to solve the closed-box nature of the first-generation; intelligent textbooks were difficult to extend, modify, and scale.
The third generation, “intelligence extracted”, involved a fundamental perspective shift with respect to the roles of textbooks and artificial intelligence. Machine learning and natural-language processing techniques were utilized to extract knowledge from textbooks at scale.
The fourth generation, “intelligence data-mined”, involved the advent of collecting, processing, and analyzing usage data.
The fifth generation, “intelligence generated”, explores the incorporation of generative artificial intelligence.
On this note, in Towards an AI-augmented Textbook, the Google LearnLM Team (2025) indicate that, using generative artificial intelligence, textbooks can be transformed, augmented, adapted, and personalized to increase content engagement and efficacy.
Their results show that there is a tremendous opportunity to reimagine textbooks in this age of generative artificial intelligence. To view their demonstrations, please click here.
To browse and participate in our discussion area, please visit here.
Bibliography
Sosnovsky, Sergey, Peter Brusilovsky, and Andrew Lan. “Intelligent textbooks.” International Journal of Artificial Intelligence in Education 35, no. 3 (2025): 967-986.
The Google LearnLM Team, Alicia Martín, Amir Globerson, Amy Wang, Anirudh Shekhawat, Anna Iurchenko, Anisha Choudhury, Avinatan Hassidim, Ayça Çakmakli, Ayelet Shasha Evron, Charlie Yang, Courtney Heldreth, Diana Akrong, Gal Elidan, Hairong Mu, Ian Li, Ido Cohen, Katherine Chou, Komal Singh, Lev Borovoi, Lidan Hackmon, Lior Belinsky, Michael Fink, Niv Efron, Preeti Singh, Rena Levitt, Shashank Agarwal, Shay Sharon, Tracey Lee-Joe, Xiaohong Hao, Yael Gold-Zamir, Yael Haramaty, Yishay Mor, Yoav Bar Sinai, and Yossi Matias. “Towards an AI-augmented textbook.” arXiv preprint arXiv:2509.13348 (2025).