AI or Advisor? Which Student Questions Each Should Answer

Sunday, 11:40 PM. A first-year texts: "when's the last day to drop?" That's a fact on the academic calendar, and an AI chatbot with a current campus resource base should answer it in seconds, with the date and where it came from. Her next text is "should i drop chem?" That's a different kind of question. The answer depends on her financial aid, her degree plan, how far behind she really is and what she wants, and it deserves a person.

The line between those two texts is what this article is about. The short version: AI should answer logistics and facts, and can help students plan and follow through with a human in reach. People should answer money decisions, health and safety, and anything that needs judgment or authority. The handoff between them should feel like one conversation, not a door closing.

Who should answer what

Sort student questions by what a wrong answer would cost, not by how hard they sound.

Notice that the dividing line isn't "easy versus hard". A question about drop deadlines is easy and an AI should take it. A question about whether to drop is also short, and it belongs to a person, because the cost of a confident wrong answer lands on the student's transcript or aid package.

Why logistics belong to AI

Every front desk on campus answers the same dozen questions every week: hours, locations, who to contact, when something is due. The answers usually exist somewhere on a website, and students still ask, because at 11 PM they can't find the page and the office is closed. An AI that answers those questions well, from a campus resource base someone keeps current, gives students an answer when they need it and gives staff back the hours they spend repeating themselves.

Three conditions make it work:

• The resource base is current. Hours change by term, break schedules change every year, and offices move. Somebody has to own the details and check them before each term.
• Answers show their source. A student who can see that the drop date came from the academic calendar trusts it, and staff can check it.
• It says "I don't know". When the details aren't in the resource base, the right answer is a plain "I'm not sure, here's who knows", with a name or an office, never a guess.

Planning and nudges, with a person in reach

The middle tier is where AI is most useful and least appreciated. Students rarely need information to start a paper or prepare for registration night. They need the task broken into a first step, a reminder at the right moment, and someone checking back. An AI that texts can do that patiently, for every student, at the hour it matters.

The boundary comes when a plan touches something with consequences. A summer class that may or may not count toward a requirement, a schedule change that affects full-time status, a plan that depends on aid arriving: the AI should say so and bring in the advisor, not settle it.

11:415G

Journey

Text Message
Sun 11:41 PM

when's the last day to drop

friday at 5 pm, per the academic calendar. after that it's a withdrawal, which shows on your transcript.

should i drop chem

that's a real decision, and it can touch your aid and your degree plan, so it deserves 15 min with your advisor before friday. want me to send the booking link? a) yes b) not yet

a

here it is. bring your current grade and one sentence on why you're thinking about it, so the 15 min goes to the decision.

ok

Text Message

Tap send for her reply

An illustrative thread with a fictional student. Journey answers the fact and hands the decision to a person.

Where people must answer

Money

An AI can explain what a credit balance is, what a hold means or what a payment plan generally involves. It shouldn't tell a student whether to withdraw, borrow or sign up for a plan, because the right answer depends on a record it shouldn't be reasoning about alone and on rules your financial aid staff know best. The AI's job is to make the student's question clear, tell them what to bring, and get them in front of the person who can see their account.

Health and safety

Here the rule is simple. Journey is not a crisis line: when a message raises a safety concern, it surfaces the right human path (counseling, 988, the on-call staff your team names) and follows the escalation process set up with your campus. Any AI that texts students has to recognize safety language, respond right away with care and a real path (in the US, the 988 Suicide and Crisis Lifeline takes calls and texts at 988, around the clock), and alert people through the process your campus sets. It never counsels and never diagnoses. In Journey, a message that shows a student in crisis gets an immediate reply with 988 and the campus's health and wellness contacts, and the campus team is alerted. A student can also text "talk", and a staff member reaches out.

There's a nuance worth keeping. "how do i make an appointment with the therapist" is a logistics question about counseling, and the most helpful thing an AI can do is answer it quickly with the booking details. A question about counseling isn't the same as a crisis, and treating every mention of the counseling center as an emergency teaches students not to ask.

Judgment and authority

Dropping a course, changing a major, petitioning for an exception, anything involving conduct: these need someone with context and the authority to decide. An AI can quote published policy and help the student get ready for the conversation. If a student asks whether something is against residence hall policy, it can point to the written policy and the residence life office. The judgment call, and its consequences, stay with people. When a student tells Journey they're leaving the university, Journey notifies staff, because that conversation belongs to a person.

How the handoff should feel

A bad handoff feels like being dropped: "I can't help with that. Please contact the advising office." A good one feels like being walked across the hall.

1. Say why. "That's a real decision, and it deserves 15 minutes with your advisor." The student should know this is a step up, not a brush-off.
2. Ask before sharing. Outside a safety concern, nothing the student wrote goes to an office without their OK.
3. Pass the context. With consent, the office gets what the student needs, so nobody tells their story twice.
4. Make booking one tap. A link that opens the right office's scheduling page, not a phone number to call during business hours.
5. Check back. After the appointment, one text: did it happen, did it help?
6. Be honest about what's AI. Students should always know when they're texting software, and when a person has taken over.

Some requests sit between an answer and a decision, like a student asking for a bus pass to get to the career fair. Approving a request like that from Journey's dashboard, with Journey texting her where to pick it up, isn't live yet. Today a request like that is flagged for staff, with alerts going to your team's Slack, and they decide and reply through their usual process.

What to measure

• Answer accuracy. Pull a random sample of AI answers every month and have the owning office check them.
• "I don't know" by topic. Each one points to a gap in the resource base you can fill.
• Repeat questions. A student who asks the same thing twice didn't get a usable answer the first time.
• Handoff to human contact. Hours from the AI saying "this needs a person" to a person actually reaching the student.
• Safety routing. Every message with safety language reached a person, and how fast. This one should be reviewed with counseling, case by case.
• Advisor time on logistics. Ask advisors at the start and end of the term how much of their week goes to questions a resource base could answer.

Questions to ask any vendor

1. Show me the list of question types your AI won't answer. Can we change it?
2. How does it decide to hand off, and what does the student see when it does?
3. Where do answers come from, who updates them, and can a student see the source?
4. What happens to a message with safety language at 2 AM, step by step?
5. Does the student always know they're texting AI?
6. What do advisors see after a handoff, and what stays private?

The useful decision is which questions you want your advisors spending their week on. Give the dining hall hours to software, keep the "should i drop chem" conversations for people, and make the walk between them short.