University of Michigan · Marsal Family School of Education

Fingers. Pencils. Screens. AI.
Learning has always run on tools.
We design which tool does what — with evidence.

That is Balanced Design for Learning — not choosing between paper and AI, but giving each the work it teaches best.

Why this lab exists

Every tool earns its place in learning

Students learn differently with different tools. Writing by hand, building with materials, and working on screens each shape understanding in their own way — and none of them replaces the others.

We design learning environments where every tool — from pencil to AI — does the work it does best.

Classrooms today hand more and more of learning to screens by default. We ask a quieter question: what does each medium do best? Deciding that with evidence — not by fashion — is how we help all students build the problem-solving skills they use in everyday life.

Research

UDL asked which learners.
We ask which medium.

Grounded in twenty years of learning-progressions research, our work spans the three places where the choice of medium shapes learning — what students work with, how they are supported, and how we know they are growing.

01 · Curriculum

Balanced learning materials

Evidence-based curriculum design that keeps hands-on work — paper, worksheets, manipulatives — where it teaches best, and prints from the same source that powers the digital pipeline.

02 · Instruction

AI as a learning partner

AI that analyzes student work, spots learning gaps, and returns time to teachers — guiding students with questions rather than answers. A partner, never a replacement for professional judgment.

03 · Assessment

Measuring where learners are

Learning progressions and AI scoring of handwritten student work, so decisions about physical and digital media rest on measurement — not on intuition or fashion.

Current project

This idea is already at work in Alabama

ColSci — Collaborative Science

U.S. Department of Education · EIR (Education Innovation and Research), early-phase · 5 years

Across low-income schools in Alabama, students work on printed, hands-on science materials — writing, drawing, and building by hand. Their work is then read by an AI assessment engine that gives teachers timely insight through a dashboard.

Digital does the authoring, scoring, and feedback. Paper sits in front of the student.

The loop: curriculum builder → printable lessons → students work on paper → AI scores the work → teacher dashboard → feedback flows back into the curriculum.

80partner schools in Alabama

6,400students

160teachers

People

Principal Investigator

Namsoo Shin, Ph.D.

Research Scientist · Marsal Family School of Education, University of Michigan

Dr. Shin researches how students learn and designs learning environments that help all students — especially in STEM — develop the problem-solving skills they use in everyday life. Her work spans conceptual frameworks, research methodologies, and technology-integrated learning environments.

For twenty years she has built and validated learning progressions — empirical maps of where learners are — the measurement backbone that lets this lab decide media questions with data.

2003University of Michigan, hi-ce — Center for Highly Interactive Computing in Education

2015Michigan State University — Associate Director of Innovation and Research, CREATE for STEM Institute

2026Returns to Michigan — Marsal Family School of Education, founding BDL Lab

Researchers

Sahana Manjunath

Researcher · AI & Systems

AI analysis of student answers, feedback generation, and uncertainty estimation for the assessment engine.

Researcher

Profile coming soon

Researcher

Profile coming soon

Researcher

Profile coming soon

Fingers. Pencils. Screens. AI.
Learning has always run on tools.
We design which tool does what — with evidence.

Every tool earns its place in learning