If AI Can Do the Assignment, What Does the Degree Still Certify?

Universities are becoming clearer about the times when students can utilize artificial intelligence. A more challenging question is arising regarding how they can demonstrate what graduates are capable of achieving without it.

For generations, the university degree has operated on a relatively straightforward assumption.

A student goes to classes, finishes assessments, passes exams and ultimately earns a qualification. This credential then accompanies the graduate into the job market as proof that a specific level of knowledge and skill has been attained. Additionally, artificial intelligence is starting to complicate this situation.

The concern has shifted from whether students are using ChatGPT or other generative AI systems to complete assignments. Universities have dedicated a significant amount of time in recent years to developing policies that address this issue.

If artificial intelligence can help research, structure, analyse, write, code and refine the work submitted for assessment, what exactly does the resulting qualification certify about the person who receives it?

That question is beginning to move from the margins of the AI debate towards the centre of higher education.

Universities Regulate AI, Face Assessment Challenges

A newly published paper, What Does the Credential Still Certify? Cognitive Stewardship for AI-Mediated Education, examined verified public generative-AI assessment guidance from 30 universities.

Universities are improving in their ability to define the boundaries of AI use, including what students can delegate, what needs to be disclosed, and what is considered unacceptable substitution.

But defining whether AI is allowed is not necessarily the same as demonstrating what a student has learned.

The researchers argue that policies remain less developed when it comes to the evidence required to preserve the validity of the credential itself.

A university can require a student to declare that AI was used to help structure an essay. It can specify whether AI-generated text is permissible. It can establish penalties when those rules are breached.

None of those measures, by themselves, necessarily establish whether the graduate can independently analyse the problem, defend the argument or apply the underlying knowledge.

The challenge thus extends beyond academic integrity and raises a question of credential integrity.

The Assignment Is No Longer the Same as the Ability

Traditionally, universities have often treated the finished piece of work as evidence of the intellectual process behind it.

A strong essay suggested the student could research, reason and write. A successful programming assignment demonstrated coding ability. A detailed report suggested analytical competence.

Generative AI weakens that assumption because increasingly sophisticated work can now be produced through collaboration between human and machine.

That does not automatically make the work illegitimate.

In many professions, using AI effectively may itself become an important graduate capability.

The difficulty lies elsewhere as universities must increasingly distinguish between the ability to produce an outcome with AI and the underlying knowledge required to understand, challenge and take responsibility for that outcome.

Those abilities are related but not identical, and employers may eventually show greater interest in the difference.

The Degree May Have to Become More Evidential

This could change assessment significantly.

Rather than relying primarily on a polished final submission, universities may increasingly need multiple forms of evidence demonstrating how students reached their conclusions.

That could mean greater use of oral defence, supervised assessment, practical demonstrations, presentations, portfolios, iterative drafts and records showing how AI was used during the process.

Another recent cross-national analysis of university AI guidance identified several emerging approaches, including portfolios documenting the learning process, AI-assisted assignments followed by oral or classroom verification, assessments requiring students to critique AI outputs, and combinations of AI-enabled coursework with secure examinations.

The direction is significant.

The future assessment system may not seek to eliminate AI from education and may instead require students to demonstrate their competence in conjunction with it.

Employers Might Pose a Comparable Question

Employers have traditionally used degrees partly as signalling mechanisms. A qualification from a recognised institution communicates that the applicant has completed a certain level of structured education.

But AI could gradually increase the importance of another question during recruitment:

What can this person actually demonstrate?

Portfolios, technical tests, case exercises, interviews, professional certifications, internships and evidence of applied experience may become more important complements to academic qualifications.

This does not necessarily diminish the value of university education.

It may instead raise expectations of what a university credential must represent.

A graduate who understands how to use AI intelligently, verify its outputs, recognise its weaknesses and still demonstrate independent judgement could become more valuable, not less.

The credential would then certify something more sophisticated than the ability to produce assignments without technological assistance.

It would represent the ability to think and perform in an environment where technological assistance is normal.

Prohibiting AI Might Address the Incorrect Issue

It is easy to view the issue as a competition between traditional education and artificial intelligence, but this perspective is ultimately too simplistic.

AI is already moving into workplaces where today's students will eventually be employed. Universities therefore face an uncomfortable balancing act.

If they prohibit the technology too aggressively, graduates may leave university poorly prepared for AI-enabled workplaces.

If they integrate it without redesigning assessment, institutions risk becoming less certain about whose capabilities they are actually measuring.

The stronger response sits somewhere between those extremes.

Students should learn to use powerful tools.

Universities must still be able to demonstrate that the humans using those tools possess the knowledge, judgement and intellectual foundations their qualifications claim to represent.

The True Worth of a Degree Lies in Trust

This may be the larger issue confronting higher education.

A university degree has never simply been a certificate documenting several years spent on campus, but a promise.

It tells employers, governments, professional bodies and society that the institution has examined an individual and is prepared to certify that certain standards have been reached.

Artificial intelligence does not automatically break that promise.

But it does require universities to reconsider how they prove it.

The institutions that respond most effectively may not be those with the strictest AI policies. They may be those that redesign assessment so that AI use and human capability can coexist without becoming indistinguishable.

In the age of artificial intelligence, the most important question may no longer be whether a student used AI to complete an assignment.

It may instead focus on whether the university can still confidently answer a much simpler question,

What can this graduate actually do