Meet MIA: Why the Future of Campus AI Depends on Verified Identity

Written by

in

AI assistants are becoming persistent.

Instead of opening a new chatbot every time they have a question, people increasingly expect digital assistants to remember context, understand ongoing projects, help manage tasks, and act on their behalf. That shift is already beginning to reach higher education, where machine intelligence tools are being explored for advising, tutoring, research, scheduling, and administrative support.

But persistence changes the problem.

A chatbot can answer a general question without knowing much about the person asking it. A persistent assistant cannot safely do the same things.

If an assistant is going to know a student’s courses, help with a research project, access institutional information, or send a message on a student’s behalf, it has to know more than what the user typed into a chat window. It has to know who the person is, which institutional relationships are current, what that person is authorized to access, and what the assistant itself is permitted to do.

That is the problem College Contact proposes to address with MIA — the Machine Intelligence Assistant.

MIA Isn’t Just a Better Chatbot

MIA is the term College Contact uses for a persistent, identity-aware tutoring and mentoring assistant for students and researchers.

The important word is not intelligence. It is identity.

A conventional chatbot can respond to a question based largely on the information contained in the conversation. An MIA would operate within a broader institutional context. Over the course of a semester, for example, it could understand a student’s active enrollment, relevant courses, institutional affiliation, and previously established context.

For a researcher, that persistent relationship could extend across a multi-year project.

That creates useful possibilities. It also creates a new category of risk.

An assistant that does not know who it is dealing with can only guess at authorization. An assistant that knows who someone is but does not know what they are currently authorized to do can still make mistakes. And an assistant that knows both but is allowed to make consequential decisions on its own creates an entirely different governance problem.

College Contact therefore treats identity, authorization, and human oversight as part of the foundation of MIA—not as features to be added later.

The Simple Request That Isn’t Simple

Consider a student telling an assistant:

“Email my professor about the extension.”

To a person who knows the student, that request might seem trivial.

For an institutional system, it isn’t.

Before acting, the system potentially needs to establish:

  • Who is making the request?
  • Which institution is the student currently affiliated with?
  • Which course does the student mean?
  • Who is currently teaching that course?
  • Is the student actually enrolled?
  • Is the instructor’s affiliation current?
  • Is the student authorized to use the relevant communication system?
  • What information can the assistant disclose?
  • Is the assistant authorized to send the message, or only prepare it?
  • Can the resulting action be audited?

A generic chatbot has no reliable way to answer many of those questions.

It can infer.

That is precisely the problem.

College Contact proposes that an MIA resolve this kind of request against verified identity, current affiliation, role, and authorization information, rather than attempting to infer institutional relationships from conversation context alone.

The difference is subtle from the student’s perspective. The student still says, “Email my professor.”

But underneath that simple request, the system is no longer guessing.

Identity Is Not Authorization

One of the most important distinctions in the College Contact proposal is that knowing someone’s identity is not the same thing as knowing what they are allowed to do.

A student can be legitimately enrolled at a university without being authorized to access every university system.

A professor can have a verified institutional affiliation without being authorized to view a particular student’s records.

A researcher can be associated with a project without being authorized to access every dataset associated with that project.

MIA therefore cannot simply ask, “Who is this person?”

It also has to ask:

“What is this person authorized to access or do right now?”

That distinction is central to the architecture proposed in the College Contact white paper.

It is also why a persistent assistant presents a fundamentally different identity challenge from a conventional chatbot.

Why the Identity Layer Matters

College Contact’s broader proposal separates the human-facing communication identity from the infrastructure responsible for verification, authorization, and auditing.

The proposed architecture includes role-oriented identities such as:

  • professor.university for verified professor-rank identities in research-university contexts
  • professor.college for verified professor-rank identities in college contexts
  • lecturer.college for verified lecturer-rank identities
  • lecturer.email for a human-facing lecturer communication identity
  • college.email for institutional college communication
  • university.contact and college.contact as public verification layers

The naming itself is not supposed to prove anything.

That distinction matters.

A domain name can tell a human what an address claims to represent. It should not, by itself, establish that the person behind that address currently holds the claimed position.

Verification and strong authentication still have to happen underneath.

In other words:

The name is a signal. The verification system is the trust mechanism.

That is the architecture an MIA could use when it needs to turn “my professor” into an actual, currently affiliated person.

MIA Should Not Become the Institution

There is another boundary that is just as important as identity.

MIA is not intended to replace professors, advisors, researchers, administrators, or academic governance.

College Contact does not propose giving MI systems authority over consequential academic or personnel decisions such as:

  • grading
  • admissions decisions
  • tenure decisions
  • academic-integrity findings
  • disciplinary proceedings
  • research supervision
  • other decisions that properly belong to human institutional governance

An MIA can assist.

It can explain material, help a student practice a concept, organize a study plan, locate information, prepare a message, help a researcher organize literature, or track project milestones.

But assistance should not quietly become authority.

That boundary is deliberately explicit because highly useful systems have a tendency to accumulate responsibility over time. If an assistant consistently gives good answers, people naturally begin trusting it with increasingly consequential tasks.

The safe response is not to make the assistant less useful.

It is to define its authority before that authority expands accidentally.

The Reserved Tier: Let MIA Learn Before Students Depend on It

This is one reason the College Contact proposal includes a deliberately separate set of reserved domains for MI testing.

The proposed beta-testing tier includes:

  • universities.email
  • colleges.email
  • lecturers.email
  • professors.college

These domains are not intended to become ordinary production addresses during the testing period.

Instead, they provide a bounded environment in which identity resolution and MI-assisted communication workflows can be evaluated before those workflows are trusted with real institutional communication.

That distinction is important.

If an experimental system incorrectly resolves an identity in a sandbox, the failure can be observed without putting an actual student, professor, or institution at risk.

The proposal also does not treat graduation from the beta tier as automatic. The white paper calls for predefined pilot metrics, independent security and privacy review, and an opportunity for participating institutions and faculty governance bodies to evaluate the results.

In other words:

MIA should have to earn the right to operate in production.

What Happens When Identity Goes Wrong?

Better identity infrastructure does not make MI infallible.

An assistant could still misinterpret a request. Institutional records can be outdated. Affiliations can change. Authorization rules can be complicated. Academic roles can be disputed.

And a system can be technically correct about identity while still being wrong about what it should do.

Consider a few possible failures:

The wrong professor.
A student has two courses with instructors who share a surname. The assistant chooses the wrong recipient.

The former affiliation.
An instructor recently left the institution, but an outdated record still identifies them as active.

The right person, wrong authorization.
A student is correctly identified but asks the assistant to retrieve information they are not entitled to access.

The right identity, excessive disclosure.
The assistant has legitimate access to a student’s information but exposes more of it than the task requires.

The correct action, insufficient oversight.
The assistant performs an action that is technically permitted but consequential enough that a human should have reviewed it first.

These aren’t simply “AI problems.”

They are identity, authorization, privacy, governance, and accountability problems.

That is why putting an identity layer underneath MIA matters.

The Data Question

There is an even bigger question that becomes possible once identity is standardized.

A national identity and communication framework could, in theory, make it easier to conduct research across institutions. Longitudinal studies could become easier. Researchers could potentially study patterns that are invisible inside a single institution. MIA systems could potentially benefit from better understanding of how students interact with educational tools over time.

But the same capability creates a significant risk.

The infrastructure designed to establish identity and authorization should not automatically become a national pipeline for collecting student and researcher data.

Those are different purposes.

College Contact therefore treats broader data aggregation for MIA development or research as a separate question requiring its own governance, consent structure, legal analysis, privacy protections, and research oversight.

That separation is important because the ability to collect data is not a justification for collecting it.

The identity layer should answer:

Who is this person, and what are they authorized to do?

It should not quietly evolve into:

What can we collect about this person because the infrastructure makes it possible?

MIA Is a Test of the Architecture

It is tempting to think of MIA as an application that could simply be bolted onto existing campus systems.

But MIA exposes the limitations of fragmented identity infrastructure unusually well.

A persistent assistant needs continuity.

Continuity requires context.

Institutional context requires trustworthy identity and authorization.

And actions taken on a person’s behalf require accountability.

That chain leads directly back to the central argument of College Contact.

MIA is not the reason to build a national higher-education identity and communication framework. It is one of the clearest tests of whether such a framework actually works.

If the underlying identity system cannot reliably establish who a person is, which institutional relationships are current, and what that person is authorized to do, then making the assistant more intelligent does not solve the fundamental problem.

It may simply make the wrong answer more convincing.

The Future Campus Assistant

The most useful campus assistant of the future may not be the one with the most impressive conversational ability.

It may be the one that can reliably answer a different set of questions before it acts:

Who am I talking to?

What do I know about their current institutional relationship?

What are they authorized to access or do?

What am I authorized to do on their behalf?

What information is actually necessary for this task?

When does a human need to remain in the loop?

Those questions are less glamorous than model capability benchmarks, but they are what separates a helpful assistant from an institutional system that can be trusted with real responsibilities.

College Contact’s proposal is that higher education should address those questions before persistent MI becomes deeply embedded in campus life.

MIA is one possible expression of that future: a persistent assistant that can know a student or researcher over time without turning that relationship into unchecked institutional authority.

The goal is not an assistant that can do everything.

It is an assistant that knows who it is helping, what it is allowed to do, what it is not allowed to do, and when it needs a human.

That is the difference between a chatbot with institutional access and a genuinely trustworthy machine intelligence assistant.

This article expands on Sections 6, “Communication Architecture,” and 7, “MI Readiness,” of the College Contact white paper.

Comments

Leave a Reply