Are student services teams accountable for an AI agent’s mistakes?

Student services advisers are accountable for the decisions they approve: extensions, fee arrangements, refusals and replies that apply policy to a student's situation. They are not accountable for factual answers the agent gave within bounds others set, or for gaps in the agent's sources they could not see. The record shows which decisions were theirs.

What does an agent handle in student services?

A university or college student services team answers enormous volumes of questions: deadlines, fees, accommodation, enrolment, wellbeing services. An agent can answer factual questions from approved policy, check a student's record, draft replies and prepare cases for decisions such as extensions, fee payment plans and withdrawal.

The decisions that matter to students are the ones that apply policy to a person. Whether a student gets more time, a fee plan or a refund changes their year. Those stay held for an adviser, and some go further, to an academic panel.

What may the agent do, and what stays with people?

• Allowed within bounds: answering factual questions from the approved policy library, confirming deadlines from the student record, booking appointments.
• Held for an adviser: any extension request, fee plan or refund, any reply that mentions health, finances or family circumstances, and any case where the student says they are struggling.
• Refused: deciding academic appeals, and sharing a student's information with a parent or third party without recorded consent.
• The adviser keeps: judging circumstances, routing wellbeing concerns to the right service, and anything that sets a precedent.

What does the record show about who decided?

Every reply and draft decision carries its sources. The student record and policy library are sealed. A student's explanation of why they need an extension is declared. A doctor's letter they upload is signed. The record shows what the agent proposed, which sources it used, what the adviser saw and what they approved.

If a student later appeals, the institution can reconstruct the decision. Public bodies, and many education providers, face extra duties of fairness and transparency, so this matters beyond the individual case.

An example: the extension refused on the rules alone

Take a university with a policy that extensions need supporting evidence. A student asks for one, mentioning a family bereavement but attaching nothing. The agent drafts a refusal citing the evidence rule, flags the mention of bereavement, and holds the reply. The adviser, clearing a peak-season queue, approves the draft.

The student fails the module and appeals. The record shows the flag, the draft and the approval. The adviser is accountable for refusing without contacting the student about evidence or support. The institution is accountable for whether advisers knew that a flagged circumstance meant "speak to the student", not "apply the rule".

What controls keep this fair?

1. Personal circumstances flags trigger contact, not a refusal. The agent should draft a reply asking for evidence and offering support.
2. No third-party disclosure without recorded consent, checked against a sealed source.
3. Advisers practise on real requests first, through a two-week practicum for student services teams recorded in a practicum log template for student services teams.
4. Review at each renewal. Agent certificates lapse after 90 days, and policies change each term; see how to prepare student services teams for 90-day recertification.

Healthcare has a close parallel; see patient services teams and AI agent mistakes. The legal sector offers another; see paralegals and AI agent mistakes. The wider topic is in AI literacy and certification for people who work with agents.