Skip to content

AI in Education

In one line: For schools and universities, AI helps most with lesson prep, feedback drafts and tutoring support, and goes wrong when it replaces thinking the student should do, exposes student records, or is used to accuse students without evidence. Last reviewed: 2026-11-07

Not legal advice. Rules differ by country, state and district; check with your institution's counsel and privacy officer.

The short version

  • Teachers and staff: strong uses are lesson and quiz drafts, differentiated reading levels, rubric-based feedback drafts, parent-communication drafts, admin summaries.
  • Students: AI tutoring helps when it asks questions and checks understanding; it hurts when it just hands over answers. Set clear rules for each assignment about what AI use is allowed.
  • Student records are protected. In the US, FERPA (20 U.S.C. 1232g; regulations at 34 CFR Part 99) gives parents, and eligible students once they turn 18 or enter a postsecondary institution, rights to access records, request amendments and control some disclosure of personally identifiable information. (US Department of Education student privacy page, read 2026-11-07)
  • The Department's page names education technology vendors as an audience, so vendors are expected to know FERPA too. It did not detail full coverage; confirm specifics in the regulations.
  • Do not rely on "AI detectors" as proof of cheating. This is a judgment: detectors are probabilistic and can be wrong, so treat them as a prompt for a conversation, not evidence. No study was fetched for this page.

Uses by role

Role Good uses Watch for
Teacher Lesson plans, examples, question banks, feedback drafts Accuracy, bias in content, student data in prompts
Student Explanations, practice questions, writing feedback Skipping the learning, undisclosed use
Administrator Policy drafts, communications, scheduling Procurement, privacy, equity of access
Tutor/tool vendor Socratic tutoring, adaptive practice Data retention, age-appropriate design

Set up in a school

  1. Write an acceptable-use policy that separates "AI allowed", "AI allowed with disclosure" and "no AI" for each kind of assignment. See acceptable use policy.
  2. Approve tools centrally. Check the vendor's data terms before any student data goes in.
  3. Teach staff and students what models get wrong. See hallucinations.
  4. Redesign assessment where needed: oral checks, in-class work, drafts with process evidence.
  5. Review outcomes each term and adjust.

Copyable prompt

I teach [subject] to [age/grade]. Write 5 practice questions on [topic] at
three difficulty levels, with a one-line hint for each that does not give
the answer. Then give a marking guide. Do not use any real student names.

Common mistakes

  • Pasting student work with names into a consumer chatbot.
  • Banning AI without changing the assessments that made cheating easy.
  • Using a detector score as a verdict.
  • Assuming every student has equal access to the tools.

Next steps

Sources

  • US Department of Education, Student Privacy: what is FERPA: https://studentprivacy.ed.gov/faq/what-ferpa (fetched 2026-11-07)
  • The Department's AI pages (ed.gov) returned HTTP 403 and are not cited.