Every Party at Once, or None at All: Digital Transformation, AI, and Who Chooses for Students
I teach technology in a public high school academy in California. Most of my students are low-income and Latine. Two AI systems are part of their school day now, and I can tell you exactly how each one got there. The first is spawn, an AI game studio I built for them this semester. In just over three weeks they made 396 games in it, and a fifth of their questions to its AI partner came outside school hours, on their own time. A tool like that normally has several parties to it, each checking the others: a company that builds it, a district that approves it, a teacher who uses it, a researcher who studies it. For spawn, I am all four. I built it, I sit on my district’s AI committee, I teach with it, and I study how my students learn with it. My district has not approved it yet. Every decision about it was made, and every one was made by me. That is every party at once. The second is Google Gemini, the only general-purpose AI chatbot my students can open on their own at school. Nobody chose it for them. It arrived with Google Workspace and was switched on in an administrator console, with no educational review. That is no party at all. On 1 January 2027 California's student-privacy law changes, and it assumes someone decides: a vendor bound by contract and a district that sets the rules. Good practice adds a reviewer who is not the builder. This paper is about what we owe students when AI enters their school day, and what happens to that obligation when the decision is made by one person, and when it is made by nobody. Governance ain't easy, but it's necessary, or we end up where we are. Proof, Not Answers asked for proof of the student and proof of the tool. This one asks who chooses.
Authors
- William C. Chesher (ORCID: https://orcid.org/0000-0003-4049-399X)
Publication Details
- Journal
- Figshare
- Published
- 2026-11-11
- DOI
- https://doi.org/10.6084/m9.figshare.34981360
- Primary Topic
- Ethics and Social Impacts of AI
- Type
- preprint