AHM Publishes Five Recommendations for Doctoral Advising in the Age of AI

AI turns doctoral advising into a shared responsibility

The Association for Human Mathematics (AHM) is putting doctoral advising practices under the microscope as AI becomes part of academic work. Its Teaching Working Group has published five recommendations for departments, advisors, and students, pairing principles with practical steps for putting them into use. The announcement doesn’t describe the recommendations individually, so their value can’t be judged by detail alone; the broader move is to treat AI use as an advising issue, not simply a matter of tool choice.

That framing makes sense. Doctoral work depends on a student’s ability to develop questions, assess evidence, and explain an argument in their own voice, while advising shapes how those abilities are learned and evaluated. When AI tools enter that process, decisions about their use can affect the working relationship between student and advisor, as well as the department’s expectations. A policy that speaks only to permitted software may miss the questions that arise in day-to-day research and writing.

Here’s the thing: a tool can help with a task without making the intellectual responsibility disappear. Students and advisors still need to decide who is accountable for claims, how assistance should be described, and where a student’s own contribution begins and ends. Those questions are difficult to settle through broad slogans. They need to be discussed in the context of actual academic work.

Authorship is more than producing sentences

AHM’s guidance focuses in part on preserving human authorship in student writing. That concern reaches beyond whether a student typed every sentence. Writing is one way doctoral students form and test their ideas: they choose what matters, connect evidence, and make decisions about how an argument should unfold. If a tool takes on more of that work, the final text may give advisors an incomplete picture of the student’s understanding.

That doesn’t mean every use of AI has the same effect. Similar systems can be used for different kinds of assistance, and the distinction between support and substitution depends on the task and the student’s role in it. A department’s guidance is more useful when it helps people talk through that distinction than when it assumes the label “AI use” describes one uniform practice. AHM says its recommendations pair principles with practical steps, a useful structure if those steps help turn abstract concerns into questions students and advisors can answer.

There’s a practical advising issue here, too. Advisors often assess a student’s progress through drafts, conversations, and revisions, and those activities can reveal how an argument is taking shape. When AI is involved, the advisor may need a clearer account of what the student did and what the tool contributed. (And this is the part most people overlook.) Clear expectations can protect the advising relationship from avoidable uncertainty, but they’ll only work if students understand them before a disagreement arises.

Limits and ethics need to be discussed together

The recommendations also call for limiting dependence on AI tools and addressing ethical questions about their use. Those ideas are related. A student who relies heavily on generated material may have fewer chances to practice the skills a doctoral program is meant to develop, while decisions about what to share, credit, or trust raise separate questions of responsibility. The source doesn’t specify which tools or ethical concerns the document addresses, so departments will need to examine the recommendations themselves before deciding how they apply locally.

Broad limits can be hard to interpret without examples. Students may need to know whether a rule applies to a particular stage of drafting, a research task, or a kind of assistance; advisors need enough clarity to give consistent guidance. Similar policy questions in academic settings often turn on context, including what work is being assessed and what the student is expected to demonstrate. AHM’s emphasis on practical steps could help departments make those expectations concrete, provided the steps leave room for discussion where a simple rule won’t fit.

The numbers here are underwhelming: five recommendations alone can’t resolve every case that will come up in doctoral work. But a short framework can still be useful if it gives departments a common starting point. If you’re working in this space, the test is whether students can use the guidance to make sound decisions before they submit work, not just whether the department can point to a document after a problem appears.

Wellbeing belongs in the conversation

AHM includes doctoral students’ wellbeing among the areas its guidance addresses. That choice broadens the discussion beyond authorship and academic integrity. Expectations around AI can shape how students experience their work, especially when they’re unsure what assistance is acceptable or how openly they can ask for help. The source doesn’t explain the specific wellbeing measures proposed, but including the subject signals that advising practices should account for students as people navigating demanding research, not only as producers of written work.

That’s a meaningful inclusion. A policy that focuses only on what students must avoid can leave them with little guidance about how to ask questions or seek support. Conversely, treating wellbeing as a reason to sidestep clear expectations wouldn’t resolve confusion either. Departments and advisors have to hold both concerns in view: students need understandable boundaries, and they need an advising environment where uncertainty can be raised early.

What this means for you depends on your role. Students can look for a shared understanding with advisors about how tools fit into their work; advisors can make expectations part of regular conversations; departments can consider whether their guidance is clear enough to apply consistently. No single document will replace those conversations.

Feedback will shape the recommendations’ usefulness

AHM is inviting readers to review the recommendations and send feedback and suggestions. That invitation matters because guidance for doctoral advising has to work across different research practices and advising relationships. A principle may sound clear in isolation yet leave people unsure what to do in a specific situation. Feedback from readers can expose those gaps and help the organization understand which practical steps need clarification.

This is more significant than it looks, though the publication itself shouldn’t be mistaken for proof that departments have changed their practices. The source describes recommendations and an invitation to comment; it doesn’t report adoption or outcomes. Their effect will depend on whether departments, advisors, and students use them to set expectations together, revisit those expectations as work changes, and discuss cases where the guidance doesn’t provide an obvious answer. A useful framework, then. The hard work is applying it.

Read the AHM recommendations on doctoral advising.