Adaptation, Environmental Behavioral Sciences
In addition to a clear policy for the course, this example offers a broader disciplinary perspective.
Course Info:
- EBS / EARTHSYS 183: "Adaptation"
- Winter 2025
- Instructor: James Holland Jones
Pedagogy:
This course policy on the use of generative AI goes beyond stating clear boundaries for the course itself. It also provides students with insight into the instructor's reasoning and broader guidance to students on the use of AI beyond the context of the course. In particular, James Holland Jones comments on the nature of Earth Systems Science as an interdisciplinary field and its relationship with how LLMs work.
Course policy
On the Use of Generative AI Tools
Consistent with Stanford’s Fundamental Standard, I expect all work you turn in to be your own. The Office of Community Standards has Generative AI Policy Guidance, and it would be smart to familiarize yourself with it. In particular, “Absent a clear statement from a course instructor, use of or consultation with generative AI shall be treated analogously to assistance from another person. In particular, using generative AI tools to substantially complete an assignment or exam (e.g. by entering exam or assignment questions) is not permitted. Students should acknowledge the use of generative AI (other than incidental use) and default to disclosing such assistance when in doubt.”
You should not use generative AI tools to write your problem sets or exams. You may get asked a question that involves using one of these tools, so it will obviously be permissible in this case (assuming proper citation, etc.), but you must still write any and all answers to questions on problem sets and exams yourself.
In general, you should be extremely cautious when using the output of Generative AI tools such as large language models. Don’t trust anything ChatGPT or any other LLM tells you. If it provides a fact for you, assume it is wrong until you have verified it from other sources. This is particularly true for citations to academic work. LLMs tend to output what have been called hallucinations—misleading or false information presented as fact—but really should just be called what they are, namely, fabrications or falsehoods.
The jury is still out, but I suspect that these tools are going to be particularly bad at answering questions in highly interdisciplinary fields that require substantial synthesis across the fields, such as the material in this class. By definition, the statements you make about novel areas of study are low probability. You should also assume that I have run all problem set and exam questions through a LLM myself to see how they perform. Let’s just say this: if you want a good grade in this class, you should just attend lecture, do the readings, take notes, and synthesize the ideas yourself. Don’t rely on a LLM, because it is unlikely to help you.