Policies
This page covers the course's generative AI use policy. For attendance, late work, and grading policy, see Assignments. For the complete set of University-wide policies (academic integrity, disability & access, Title IX, and more), see the full syllabus.
Course Artificial Intelligence Policy
In this course, generative AI use is partially permitted. This is a course about language models, and students are expected to use them. By default, generative AI tools are allowed for coursework as long as you review their output critically, remain responsible for what you submit, and disclose use that contributed to the submitted work. There are deliberate exceptions, listed below, to cover the parts of the course where a student's own judgment is what is being taught and assessed. Any assignment-specific restrictions will be stated in writing.
| Activity | AI tools | Notes |
|---|---|---|
| Debugging, environment setup, boilerplate training/eval scripts | Permitted | Disclose in your AI use statement |
| Explaining concepts, papers, or error messages to yourself | Permitted | No disclosure required |
| Drafting, editing, and polishing prose in reports | Permitted | Disclose; you own every claim |
| Generating synthetic training data where the assignment calls for it | Permitted | Describe prompts and filtering in your write-up |
| Choosing your project task and domain | Not permitted | Your team must choose the task and domain yourselves. AI may be used in the brainstorming stage, but the final decision needs to be made by humans. |
| Authoring evaluation prompts, rubrics, and gold labels | Not permitted | You or your team write these; AI may critique a draft you wrote |
| Selecting and labeling preference pairs (HW2) | Not permitted | Human judgment is the point of the assignment |
| Interpreting your training runs and writing the analysis | Not permitted | The interpretation and conclusions must be yours |
| Concept check | Not permitted | Closed-device, handwritten, in class |
If a situation is not covered above, ask the instructor.
Disclosure Policy
Every graded submission must include a short AI Use Statement at the end of the document. Describe AI use that contributed directly to the submitted work; routine studying or concept clarification does not need to be listed. The statement should include:
- Which tools you used (e.g., Claude, ChatGPT, GitHub Copilot, Cursor, Codex)
- What you used them for, by task (e.g., “debugging the SLURM submission script”)
- Which parts of the submission were substantially AI-generated and how you verified them
Two or three sentences is usually enough. If you used no AI tools in preparing the submission, say so. For group submissions, each member writes a separate statement.
Failing to disclose AI use when this policy requires disclosure may constitute academic misconduct under UT Austin's Institutional Rules and may be referred to Student Conduct and Academic Integrity in the Office of the Dean of Students.
Do not upload other students' work, unpublished course materials, or any non-public data to external AI services.
University Policies
Academic integrity, collaboration expectations, classroom recording, sharing of course materials, disability & access accommodations, religious holy days, Title IX disclosure, campus safety, and the land acknowledgment are all stated in full in the official syllabus linked above.