Prompt Engineering for Education
Universities, schools, districts, and education researchers are moving fast on LLMs — not just as classroom novelties, but as production tools for tutoring, grading, differentiation, and lesson design. Teachers who can build and evaluate these systems, not just blindly use them, are increasingly in demand.
Where this is showing up in Education
- Right here at Furman! Professor Johnson, Dr. Alvin, and Tram Le are researching using LLM-based agents to create differentiated curricula using a "Curriculum-as-Code" methodology with PathMX (PathMX also powers this course site.)
- Khan Academy's Khanmigo (Socratic AI tutor, free for U.S. teachers) and MagicSchool AI (80+ teacher tools including an IEP generator) are now deployed across major U.S. districts.
- HMH's Writable won the 2024 CODiE for Best AI Implementation in EdTech — rubric-aligned feedback on student writing across grades 3-12, now bundled with HMH curricula.
- Stanford's Tutor CoPilot (Wang, Ribeiro, Robinson, Loeb, Demszky, Oct 2024) ran the first RCT of a Human-AI tutoring system: 900 tutors, 1,800 K-12 students, +4pp mastery overall and +9pp for students of lower-rated tutors, at $20/tutor/year.
- FERPA, COPPA, and state-level AI-in-schools guidance (e.g., the U.S. Department of Education's AI and the Future of Teaching and Learning report) are actively shaping what tools can ship in classrooms.
Projects you could build in this course
- A Socratic tutoring agent that refuses to give away answers and adapts to a student's prior turns
- A rubric-grounded writing-feedback tool that cites specific lines and criteria
- A lesson-plan generator bound to a state standard or curriculum framework