Department AI Guidance in One Quantitative Field
A department-wide AI guidance document provides explanations and clear baseline policies in a quantitative field.
Course Info:
- Draft Departmental Guidance in One Quantitative Field
- Summer 2025
- Departmental teaching committee
Pedagogy:
A group of faculty from a department came together during summer 2025 to draft department-wide guidance about AI use. The guidance addresses the purposes and forms of learning that are paramount in this field, consequences of over-reliance on AI, and what is explicitly prohibited. In this framework, individual instructors may still set policies for particular assignments that incorporate AI in specific ways. This departmental guidance also expresses how students are responsible for understanding the work they submit, with a connection to post-graduate expectations in the workforce.
One Example of Departmental AI Guidance (to help inform course syllabi)
The purpose of education:
We urge you to think for yourself on assignments. Thinking is hard; it is how you learn and make progress. You may consult lecture notes, textbooks, teaching staff during office hours, CTL tutors, and your peers (preferably after trying an assigned task by yourself). Sensible use of AI tools (more on this below) can devise new examples to explore and probe your grasp of ideas. However, you must write homework submissions in your own words, expressing your own thoughts, after consultations are done.
Expressing ideas and methods in your own words is how to judge what you truly understand, and real understanding is what makes a Stanford degree worth more than the paper on which it is printed. Just as watching cooking shows while sitting on a couch does not actually give you the ability to cook, if you rely on AI to do your academic work and merely read (and perhaps memorize) its output then you will never learn how to “do the math”: to use course content in substantive reliable ways in the future.
Avoid over-reliance on AI:
Using AI to do your work will prevent you from developing the problem-solving skills and conceptual understanding needed to succeed in this course and any future quantitative courses, which will lead to poor performance on exams. Course material builds on itself, and gaps in your understanding will accumulate over time. There is no substitute for your own hard work.
The only way to truly learn is to engage actively with the material and work through assignments on your own. If you use AI to generate practice material for yourself, beware that AI can be wrong. When you are stuck, ask instructors or teaching assistants for help. Unlike AI, which often provides responses without helping you to genuinely learn the material, the course teaching team will guide you through the thought process in a way that strengthens your understanding.
It is prohibited without explicit instructor permission to share, upload, or reproduce copyrighted course materials – including lecture notes, readings, homework assignments, practice exams, and exams – on AI platforms (or any websites) apart from asking it to explain or illustrate a screenshot of a limited part of the material (e.g., for clarification, learning purposes, asking AI if your own work has errors, or illustrating similar – not identical – examples). Since AI output is generated by probability rather than intelligence, it is entirely your responsibility to judge its correctness via your own intelligence. The blind-faith excuse “but AI told me this is correct” is never acceptable.
AI Disclosure Policy:
The purpose of homework assignments is to reinforce your learning and prepare you for exams to assess mastery of the course material. Apart from circumstances when the instructor explicitly permits AI to be used in specific coursework, if you outsource to AI any part of your work on an assigned task (to get ideas, to try out a similar but not identical task, etc.) then you must abide by the following AI Disclosure Policy:
- Only use the Stanford AI Playground (this is behind a Stanford firewall, and nothing you enter into it is shared outside Stanford or used for any model training, according to this). Using AI to generate solutions for homework, a project, or other assigned work is prohibited.
- Write and submit your own solution without looking at the AI output. Submitting AI-generated text as your own work is an act of plagiarism insofar as it involves passing off the work of others as your own. On exams you will be expected to demonstrate your understanding without AI tools.
- Below your solution, thoroughly document your AI use: name of the AI platform(s) and screenshots of the prompt(s) you entered, and the entirety of the AI output.
If you are unsure about policies regarding the use of AI tools in this course, seek clarification from the instructor(s). Students have been found responsible for Honor Code violations in recent years when their homework and/or exam answers clearly resemble AI output and are distinct from their usual style of writing.
Responsibility for understanding:
You are expected to understand your submitted work in homework and exams and to be able to verbally explain to your instructor (upon request, and without advance notice) any work you submit. When you apply for jobs, you will be interviewed by professionals who instantly recognize a lack of real understanding due to over-reliance on AI; we hold you to the same standard. Do not submit what you do not understand. The response “I don’t understand what I was writing but I know it works” is never acceptable when asked to explain a submission.
If your verbal explanation is inadequate in that you clearly do not understand what you wrote down or you cannot carry out a similar task using related ideas, your grade for that portion of the assignment or exam will be assigned (or reassigned) to be 0. Your instructor may also file an Honor Code concern if your work contains solutions resembling AI output that you cannot adequately explain in-person in your own words when asked to do so.