Using AI on Coursework at HWS: What’s Actually Allowed
The question every student has before touching any of this. The honest answer is that it depends on your professor — but there is a lot more to say than that, and knowing where the lines usually fall makes the conversation with your instructor much shorter.
The HWS AI Club does not set academic policy. Nothing on this page overrides your syllabus or your instructor. Where the two disagree, your instructor is right.
AI and academic integrity at HWS
Is using AI for coursework allowed at HWS?
There is no single answer, and anyone who gives you one is guessing. AI policy at Hobart and William Smith Colleges is set by the instructor, course by course. Some syllabi encourage AI for brainstorming and revision, some allow it with disclosure, some prohibit it for graded work entirely. The syllabus is the authority; when it is silent, ask before you use it.
What counts as an acceptable use?
As a rule of thumb, uses where the thinking stays yours: having a concept explained until it clicks, being quizzed on terms, getting feedback on a draft you wrote, turning a syllabus into a study plan. These are the same things a tutor or study group would do, and they leave the work — and the understanding — with you.
What counts as academic dishonesty?
Submitting AI-generated work as your own is plagiarism at essentially every institution, and it does not stop being plagiarism because a machine wrote it rather than a person. The line most policies draw is authorship: if the ideas, argument, and words handed in are not yours, you have crossed it, whatever tool produced them.
Do I have to disclose that I used AI?
Often yes, and increasingly it is the default expectation. Some courses require a note on what you used and how; some require nothing. Disclosure costs you very little and removes the ambiguity entirely, so when the syllabus does not specify, disclosing is the safer of the two mistakes to make.
Can professors detect AI writing?
AI-detection tools are unreliable in both directions — they miss real AI text and they flag human writing, disproportionately from multilingual writers. That is an argument for not relying on them, not an argument that using AI dishonestly is safe. Instructors also notice work that does not sound like the student who wrote everything else.
How does the club's use-case library handle this?
Every one of the 840 use cases is written as a study aid rather than a substitute for doing the work, and each is rated by how much checking the output needs: Easy is usable as-is, Medium should be verified against your course material, and Hard should be reviewed with a professor or TA before it goes near graded work. Every major page carries the same warning to check the syllabus first.