Key Concepts in Designing AI-Based Learning Strategies

• By Kate Lucariello
• 06/02/23

As part of the recently released "Artificial Intelligence and the Future of Teaching and Learning: Insights and Recommendations," by the Office of Educational Technology (OET) of the U.S. Department of Education (ED), OET hosted listening sessions, in which marked concerns were raised in the area of learning. Noting that ED's ed tech vision sees students as "active learners" who "participate in discussions that advance their understanding, use visualizations and simulations to explain concepts,… and leverage helpful scaffolding and timely feedback as they learn," OET heard that attendees "want technology to align and build on these and other research-based understandings of how people learn." An especially strong request was for students who have been impacted by the pandemic to be strengthened and supported in their learning.

A key insight for AI-based learning is that it enables adaptivity, with technology's "ability to meet students where they are, build on their strengths, and grow their knowledge and skills." But the caveat is that adaptivity can sometimes be too specific and limited, and important parts of learning can be left out or insufficiently developed. An important goal is to work "toward AI models that fit the fullness of the visions for learning — and avoid limiting learning to what AI can currently model well."

Even advances in "large language models" have limits, the report noted, with experts in the listening sessions warning that AI models are narrower, learning contexts can change; "common sense" human judgments are lacking; and all of these can result in "unnatural or incorrect" AI responses. New AI designs must account for that.

While intelligent tutoring systems (ITS) have advanced significantly in giving students feedback, it is still important for human teachers to motivate students and help them self-regulate, the report said, adding that "... any teacher knows there is more to supporting learning than adjusting the difficulty and sequence of materials." Human teachers have a better understanding of the totality of their students than most ed tech does and can recognize a "teachable moment" in a way that an AI cannot.

Recommendations for expanding the core models of an AI system are to:

• Design adaptivity that is "asset-oriented," building on student competencies, rather than "deficit-based," or lack-focused;

• Include social and other aspects of learning in addition to individual cognition-based learning;

• Include multiple learning strategies for neurodiverse learners and those with disabilities;

• Design models that include "active, open, and creative tasks" and innovative approaches in addition to fixed tasks;

• Expand beyond only "correct" or "incorrect" answers to teach students how to keep working on problems and ask for help when needed.

The report notes that "two broad perspectives" arise around AI in education: "AI in support of student learning" and "support for learning about AI and related technologies." On the latter, the report emphasizes the importance of students becoming educated about AI, not only about what it can do but what risks it poses.

The report applauds research and development attempts to address the recommendations for AI learning strategies and reiterates that "our key recommendation is to tease out the strengths and limitations of AI models inside forthcoming ed tech products and to focus on AI models that align closely to desired visions of learning. AI is now advancing rapidly, and we should differentiate between products that have simple AI-like features inside and products that have more sophisticated AI models."

Visit this page to read and download a summary handout of the report's main points. A webinar going into more depth on this report will be held Tuesday, June 13, 2023, at 2:30 p.m. ET. Signup is available by QR code at this link.

The full report can be downloaded from this page.