Verification Gap
In one sentence
The Verification Gap is the widening distance, in an AI-saturated environment, between what a student or job candidate can show — on a C.V., for example — and what an institution or employer can reliably know the person can do under real constraints.
Where the name comes from
AACSB used the phrase in its April 2026 employability work, particularly in the article Bridging the AI Employability Gap. The setting was business education, but the term is larger than business schools. It names a general evidentiary problem: once generative AI can help produce polished slides, polished memos, polished market scans, polished cover letters, polished portfolios, and polished interview scripts, polish stops carrying the evidentiary weight it used to carry.
This does not mean the artifacts become worthless. A clean recommendation deck still matters. A good portfolio still matters. A well-written memo still matters. But the artifact no longer proves, by itself, that the person who submitted it can frame the problem, manage the tradeoffs, validate the claims, or defend the recommendation. The output has become easier to produce than the competence the output used to signal.
That distance is the gap.
Why business schools should care
Business schools are credentialing institutions. We tell employers, graduate programs, parents, donors, and students themselves that a degree from this place means the graduate can do certain kinds of work. The trouble is that many of the traditional ways we have documented that claim were artifact-based. We collected reports, presentations, exams, portfolios, case analyses, capstone decks. We graded the artifact and treated the artifact as evidence of learning.
That practice still works when the artifact is tightly coupled to the student’s own cognition. It works less well when a student can route a vague intention through an AI system and receive, ten seconds later, an artifact that resembles the work product of a competent analyst. The resemblance may be useful. It may even be a good starting point. But resemblance is not competence.
The Verification Gap is therefore not primarily a cheating problem. Cheating is the dramatic version. The quieter problem is assurance of learning. If the artifact is no longer enough, the assessment system needs additional evidence.
What better evidence looks like
AACSB’s response is not prohibition. It is verification. The school should not pretend students will work without AI in an AI-embedded economy. It should instead ask for evidence that the student can use AI responsibly, judge its output, and own the final decision.
Useful evidence includes:
- Process notes showing how the student framed the problem and where AI entered the workflow.
- AI-use disclosures identifying prompts, tools, and the student’s edits or rejections.
- Validation memos explaining how claims, sources, calculations, and assumptions were checked.
- Oral defenses in which students answer questions about their recommendation under pressure.
- Judgment defenses requiring students to explain why one tradeoff was chosen over another.
- Version histories or worklogs that show the path from first framing to final artifact.
None of these is perfect. A process note can be faked. An oral defense can reward verbal confidence. A disclosure appendix can become bureaucratic theatre. But together they move the assessment from show me the artifact toward show me the thinking and defend the result. That is the right direction.
The employer side of the gap
The same problem appears in hiring. Employers still need to know whether a candidate can do the work. The old screening signals — resume polish, cover-letter fluency, portfolio finish, interview preparation — have been weakened. AI does not eliminate the need for talent; it makes the visible surface of talent easier to imitate.
This shifts the employer’s problem from evaluation of artifacts to verification of competence. Case interviews, work simulations, live problem-framing exercises, reference checks, and probationary project work all become more important. The credentialing institution that understands this shift can help its students. The institution that does not understand it will continue sending employers polished artifacts while employers quietly build their own verification filters.
The teaching implication
The humane response is not to make students produce uglier work so we can be sure it is theirs. That would be perverse. The response is to teach students that polished AI-assisted work is the beginning of professional responsibility, not the end of it.
A student who uses AI to draft a market analysis should be able to say: here is what I asked for, here is what the system gave me, here is what was wrong, here is what I checked, here is what I changed, here is the assumption I still worry about, and here is why I stand behind the recommendation anyway. That is competence. The final memo is only one piece of the evidence.
See also
AI Writing, Earned Parallelism, FERPA Compliance Posture, AI Produced Artifact, Institutional Lag.
Source
AACSB Insights, Bridging the AI Employability Gap, April 22, 2026; discussed in Prof. Langenkamp’s AI in Higher Education — Weekly Brief, Vol. 17, May 15, 2026.