Scott FaustAI, learning & software

Scott Faust / Researcher & builder

How can AI help us understand learning?

I study how AI models learning, what its predictions can tell us, and what evidence we need before using them to make educational decisions. I’m finishing my M.S. in Computer Science and beginning my Ph.D. in Computing & Information Science.

Explore my research A little about me
Portrait of Scott Faust
Scott FaustOmaha, Nebraska
Education
M.S. final semester
Research
Ph.D. first semester
Scholarship
First-author manuscript submitted
Foundation
U.S. Air Force veteran

Current chapter / Fall 2026

Finishing one chapter. Starting the next.

My M.S. thesis asks how learner models can make their state and evidence inspectable. As I begin doctoral study, I’m carrying that question into a broader interest in AI, assessment, and how people use learning technologies.

Questions guiding my research

Selected work

My thesis and assessment review anchor my research direction: understanding learner models and the evidence needed to use them in educational decisions.

Master’s thesis · 2026

Auditable knowledge tracing with a temporal student knowledge map

A leakage-safe knowledge-tracing system that separates shared structure, mutable learner state, and immutable evidence. The work combines temporal and graph-aware modeling with prospective experiment controls and release verification.

Step inside the learner map

A real 1,000-interaction derived-state replay. Read the thesis summary.

Problem
Predictive systems can hide state, provenance, and update logic.
Contribution
An inspectable representation and evidence-controlled evaluation workflow.
Submitted manuscript · 2026

From Prediction to Educational Consequence

First and corresponding author of a structured critical narrative review submitted to Studies in Educational Evaluation. The paper connects AI performance claims to the distinct evidence required for educational decisions.

Review the contribution
Framework
Six layers from construct definition through governance.
Boundary
Learner-state inference is separated from adaptive policy.

Research direction

Move carefully from model output to human consequence.

My research asks how representations of people, knowledge, time, and context can remain useful without concealing uncertainty or overstating what a prediction proves.

01 / Learners

Knowledge tracing

Model changing learner state while preserving temporal order, provenance, and clear boundaries around mastery claims.

02 / Decisions

Adaptive assessment

Separate what a system predicts from the action it selects and from evidence that the action improves learning.

03 / Systems

Human-centered AI

Design interfaces, controls, and audit trails that let people inspect assumptions and challenge automated behavior.

Read about current research and writing

For students / Python

Learn recursion, one call at a time.

Explore Tower of Hanoi, a 2D maze, and Sudoku with a debugger-style lesson. Step through the Python code, inspect the call stack, and watch variables change at your own pace.

Open the recursion explorer

Working principles

The way I work.

Evidence

Make claims traceable.

Separate demonstrated behavior from aspiration and preserve the evidence needed to reproduce a decision.

Systems

Design for state and failure.

Model transitions deliberately, account for invalid inputs, and treat recovery as a first-class requirement.

People

Keep human authority visible.

Build tools that support judgment without disguising model output as certainty or policy as inference.

Collaborate

Let’s talk about learning, AI, or building something thoughtful.

I welcome conversations about educational technology, learner modeling, responsible AI evaluation, and software for real operational problems.

Contact ScottView résumé