Digital Projects

An open educational resource for introductory logic, built to replace the commercial textbook in PHIL 151 at UMW and to be adoptable by other Virginia institutions and beyond. The book is drafted in full — four parts, fifteen chapters, four appendices — and was piloted in a credit-bearing course in Summer 2026. Accessibility is a hard requirement rather than a finishing touch: WCAG 2.1 AA throughout, in both the text and the tools.

The book is licensed CC BY-SA 4.0 and the tools MIT, so any instructor can adopt, remix, or self-host them. Project funding is being sought through the VIVA Open Creation Grant.

View the repository ↗ (opens in a new tab)

An AI speech coach built for my Aesthetics course. Students give a “Theory & Object” presentation — a persuasive talk connecting a physical object to theory read in the course — and are required to show evidence of a practice session beforehand. The university's Speaking & Writing Center is the human option, but its hours don't reach everyone, particularly commuter students. This tool is the always-available alternative.

Students record their presentation, upload the audio, and receive narrative feedback as a PDF. Three design decisions matter more than the code. The model processes the audio directly rather than working from a transcript, so it can hear cadence, pauses, monotone delivery, and filler words instead of inferring them from text. The actual assignment rubric serves as the system instruction, which means the feedback is answerable to the same standard the presentation will be graded against rather than dispensing generic speaking advice. And the output is deliberately qualitative — narrative coaching, no scores or numbers, nothing grade-like — because the tool is for practice, not assessment.

It was built through conversational prompting with an AI model over several sessions rather than written line by line, which is the subject of the talk below. Across the semester it logged more than twenty-five submissions, with several students returning to iterate on feedback they had received.

Reception was genuinely mixed, and I think that matters to report. Some students found it useful and said so; commuters especially. Others declined to use it, on grounds I take to be legitimate design constraints rather than noise: they did not want a machine to have their voice, or objected to the environmental cost of inference, the labor practices behind model training, or the risk of their own skills atrophying. One thought the whole thing was out of touch in an art-theory course. The tool was optional from the start and stayed that way, with the Speaking & Writing Center alongside it.

View the source ↗ (opens in a new tab) Live app ↗ (opens in a new tab)

The live app is password-protected for enrolled students; the source is public.

Reading tutors I build for my own courses, each tied to a single difficult primary text. The point is not to summarize the reading — it is to give students something that will push back while they work through it, at whatever hour they are actually reading. Students are told to start the text first and bring the bot in partway, not to use it instead of reading.

The Hegel bots work differently from the others: rather than explaining a position, they occupy one, and students find its limits by pressing on it. Each AI exercise is paired with a short piece of writing done without AI, on what the encounter did and did not demonstrate.