AI Reader
Providing students with alternative learning paths outside of the classroom
Overview
Over 50% of post-secondary students take online courses, and when they get stuck on difficult concepts at home, they're often on their own. Teachers aren't always available, and every home learning environment is different. Students often turn to tools like ChatGPT, which prevents them from developing the critical thinking skills they need to succeed.
McGraw Hill saw potential to shape how students learn outside of the classroom. Working with the AI product owner, I pushed conversations forward on what ethical, learner-focused AI tools could be, resulting in the release of AI Reader in SmartBook and more importantly, giving students the support they were seeking when they hit learning roadblocks.
Role
Product Designer
Duration
6 months
Empathize
Remember
Students naturally take the shortest path when they're stuck. At home, without teachers or parents available to guide them through difficult concepts, they turn to ChatGPT or internet searches to get unstuck and move on.
From a learning science perspective, this shortcuts learning because students who rely on quick answers get stuck at the lowest level of Bloom's taxonomy. They might achieve basic recall for homework, but they're unprepared to analyze or apply those same concepts.
Understand
In mid-2024, McGraw Hill launched AI Reader, a tool embedded directly within eBook experiences to provide additional learning paths that help students understand concepts while studying at home. With AI Reader, students can highlight text and ask for alternative explanations, simpler language, or test their knowledge with quick quizzes.
AI Reader in eBook
Apply and Analyze
Student usage was high, and McGraw Hill wanted to expand AI Reader into SmartBook, an enhanced eBook experience with adaptive modules that quiz students to determine which learning concepts (yellow and blue highlights) they should focus on.
SmartBook Highlighted Concepts (Left), SmartBook Adaptive Questions (Right)
With the addition of concepts and questions,
Iteration
Given the open-ended nature of this integration challenge, I established a continuous feedback loop, testing everything from high-level placement decisions to specific AI Reader features.
Finding Smartbook's Happy Path
Students using SmartBook follow a specific learning path: they read content, answer embedded questions, and are guided through learning concepts they don't feel confident about. When students need extra help, AI Reader found a natural home alongside SmartBook's adaptive feedback once a student gets a question wrong.
Opportunity
Supporting Without Shortcuts
We tried placing AI Reader directly within a question, side-by-side to a question, and even turning AI Reader into an AI prompt.
AI Reader directly within a question made sense from a user perspective and was a huge success with stakeholders.
However, after observing students interact with a prototype and gathering opinions from instructors, we noticed two glaring issues:
Authored content can and often is entirely skipped. This runs the risk of allowing students to learn hallucinations first, written content second.
Students more often than not end up seeking the easy path (AI), rather than first trying to grasp a concept.
For students to learn most effectively,
AI needed to be a second option, not a default. Rather than integrate AI Reader directly into the question flow, AI Reader stayed only in the eBook.
Digging deeper through talks with authors, teachers, and students, three things were clear:
SmartBook's main purpose is to help student's learn concepts. The highlighted sections are most important.
Giving students answers would immediately break teachers' trust in the tool.
Students already find SmartBook lengthy. Knowledge checks need to be obviously optional.
At first, we transformed AI Reader into a popup for each concept.
But we lost persistent history, flexibility with non-concept selections, positive mobile experiences, and hiding AI reader in the toolbar made it almost non-existent when tested.
One standout feature that users overwhelmingly found useful was the direct AI Reader button next to concepts.
With AI Reader integrated directly within a concept, students found easy yet optional access to another tool that could help them be successful outside of the classroom.
We achieved our key goals with students and instructors. Students found AI Reader seamlessly integrated into their SmartBook experience and non-disruptive to their learning flow. More importantly, students using AI Reader were able to grasp difficult concepts faster, effectively shortening their study time.
Beyond a Single Product
Students don't just learn through eBooks – they watch lectures, review notes, and study for exams. How might we support overall learning?
AI Reader is successful in its focused and limited environments, but learners do much more than just read eBooks. How could we offer the same learning benefits of AI Reader for all learning mediums?
Discovering Value
Student Learning Behaviors
First, we took stock of what students were already doing across our platform, and what could bring them value.
McGraw Hill had multiple solutions and study materials like Sharpen and AI Reader, so we looked into how others created centralized spaces for tools available throughout a platform.
We took inspiration from Google's launchpad and Notion's AI bookmarking tool. Then, we explored how Quizlet generates AI focused study materials.
Now, we just needed to combine these ideas to support our happy path.
Bringing CATs to Life
Getting Content Generated Seamlessly
From the wireframe explorations, we found that an always present, dedicated CATs button was a clear winner with students and fitting into our existing platform.
Generate and Study Shelf Menu, Bottom Right
A study space that brought together all their AI-assisted learning, regardless of where it was created was needed.
Reflection
This project pushed me to work with minimal, blue-sky requirements, which was possible because of the trust between the AI Product Owner, Learning Scientist, existing designers, developers, and myself to rely on each other's expertise.
Another takeaway was from the positive foundation we had. We weren't fixing broken products or solving major usability issues. Instead, we were enhancing already solid learning experiences with AI capabilities. A "making good products even better" mindset created space for genuine innovation and experimentation, something I want to carry into future projects.