Towards an AI-Augmented Textbook

TL;DR

The paper addresses the rigidity of traditional textbooks by proposing an AI-augmented framework that personalizes content and provides multimodal representations while preserving core material. It introduces Learn Your Way, a two-stage Gen-AI pipeline that first personalizes text (to grade level and learner interests) and then generates multiple representations and formative assessments, culminating in an immersive, self-regulated learning experience. Pedagogical expert evaluations across diverse OpenStax materials yield high-quality assessments of the components, while a randomized controlled trial with 60 adolescents demonstrates superior immediate and retention learning outcomes compared to a Digital Reader. The work signals the feasibility and value of scalable, AI-driven textbook augmentation and points to future integration into learning platforms with deeper adaptation and broader topic coverage.

Abstract

Textbooks are a cornerstone of education, but they have a fundamental limitation: they are a one-size-fits-all medium. Any new material or alternative representation requires arduous human effort, so that textbooks cannot be adapted in a scalable manner. We present an approach for transforming and augmenting textbooks using generative AI, adding layers of multiple representations and personalization while maintaining content integrity and quality. We refer to the system built with this approach as Learn Your Way. We report pedagogical evaluations of the different transformations and augmentations, and present the results of a a randomized control trial, highlighting the advantages of learning with Learn Your Way over regular textbook usage.

Figures (10)

• Figure 1: An example of the Learn Your Way learning experience. Centerpiece is the "Immersive Text" view, that shows the source material (OpenStax's Disruptions in the Immune System content) transformed to 6th grade level and adapted for a personal interest in gaming. The Immersive Text contains various generative add-ons such as personalized examples, embedded questions, and more. At any given time, the learner can also switch to alternative views of the entire material such as narrated slides or an audio lesson, which are also personalized.
• Figure 2: An illustration of the two step generation procedure used in Learn Your Way. Here, a generic example from OpenStax's Newton's Third Law of Motion is first personalized for the personal interest of 'basketball' (left) and 'art' (right), and then expanded into different presentation formats.
• Figure 3: Example slide in the deck generated for OpenStax's How To Organize Economies source and adapted to the learner interest in ‘soccer’.
• Figure 4: An example mind map created for OpenStax's Early Human Evolution and Migration source material. Different nodes can be expanded to gain a more granular view of the material, with leaf nodes annotated with text or relevant visuals.
• Figure 5: An example of personalized visual generation that captures a key concept in an engaging manner that is based on the interest of the learner in soccer for OpenStax's How to Organize Economies material. The personalized text next to the image shows that the original text has been adapted to the topic of sports. The image is a visual illustration of the text, and thus is also personalized.