Leveraging LLMs for student feedback in introductory data science courses
summary
An RStudio workflow hooks student Quarto submissions into a hosted model via RAG with semantic search and an R package front end, producing immediate, rubric-grounded feedback inside the IDE; deployed as a Shiny app and RStudio add-in tied to Canvas and GitHub repos for a 300‑student course. Feedback matches instructor rubrics about 90% of the time, reduces TA grading load, logs usage (adoption low so far), and still shows occasional hallucinations and inconsistency.
AI-written summary · about
chapters (8)
- 0:00Introduction
- 3:21First AI project
- 6:43Motivations
- 9:13Technical design
- 12:26Demo and workflow
- 13:50Implementation challenges
- 15:57Student feedback
- 17:00Next steps
related talks
11mCoGrader: Transforming Instructors' Assessment of Project Reports through Collaborative LLM Integ...
40mInstructLab: Applying Open Source Methods to Building and Trai...
44mPractical LLM Evals: Down-to-Earth Tooling for Everyone
28mA community based open source LLM and what it can do for your business- 40mInstructLab: Applying Open Source Methods to Building and Trai...
01h 34mHow to Construct and Assess Your First LLM Based RAG System
15mLLMs for Data-Driven Decision Making in Product Development
38mEmpowering AI Education with Rag: Building Domain-Specific LLM Agents for Enhanced...