A polished submission can show what a student produced. It may not show how they chose an approach, connected evidence, rejected an alternative, or responded when the answer was uncertain.
That gap is not entirely new, but generative AI has made it harder to ignore. The practical question for instructors is no longer simply, “Did a student use AI?” It is, “What evidence would let this student demonstrate the learning I actually need to assess?”
Asynchronous oral assessment offers one answer. A student records a short explanation, defense, critique, reflection, or demonstration on their own schedule. The response is anchored to a specific prompt or piece of work, and the instructor can add a focused follow-up when another turn would reveal something useful.
This is not a claim that video catches cheating. It is a way to collect another kind of evidence without scheduling a live oral exam with every student.
Table of Contents
Assessment needs redesign, not another detection layer
The Tertiary Education Quality and Standards Agency (TEQSA) describes generative AI as an intensifier of existing assessment challenges, not a problem that can be separated from assessment design. Its guidance asks higher education to preserve learning assurance while making responsible use of the opportunities AI creates.
That is a more productive starting point than trying to identify every AI-assisted sentence after submission. In 2023, Vanderbilt University disabled Turnitin's AI detector, citing concerns that included false positives, limited transparency, bias, privacy, and the underlying difficulty of detection. Vanderbilt instead recommended clear expectations, communication, and assignment redesign.
Oral assessment is receiving renewed attention in that redesign. Associated Press reporting from 2026 describes universities revisiting oral exams and defenses as instructors look for more direct evidence of understanding. Live conversation can be valuable, but it also creates a basic scaling problem: every student needs an appointment with a qualified examiner.
Asynchronous oral assessment does not recreate a live exam. It offers a smaller intervention: ask students to explain the part of their thinking that the final artifact does not make visible.
What counts as asynchronous oral assessment?
An asynchronous oral assessment has four characteristics:
- The student responds orally through audio or video.
- The response is anchored to a prompt, artifact, problem, or decision.
- The student and instructor do not need to be present at the same time.
- The design may include a later question, critique, or feedback turn.
The format is broader than a recorded presentation. A presentation often asks a student to prepare and deliver a complete account of a topic. An asynchronous oral assessment can be much smaller: explain one choice, apply one concept, critique one source, or show what changed after feedback.
It also differs from remote proctoring. Proctoring tries to control the conditions around an assessment. An oral explanation adds another mode of evidence. Those approaches solve different problems.
Cornell's Center for Teaching Innovation includes oral defenses, discussions about a paper or project, demonstrations, and short video or audio assignments within the wider oral-assessment family. Cornell also recommends pairing oral and written work when the combination better represents the intended learning outcomes.
What oral explanation can—and cannot—show
A well-designed response can add evidence that a student can:
- explain a consequential decision
- connect a claim to evidence
- apply a concept in a new situation
- critique an interpretation or counterexample
- identify the reason for a revision
- respond to a focused question
It cannot prove that the student never used AI, wrote every part of an associated submission alone, or mastered a topic simply by uploading a video. A face on camera is not identity verification, and confidence on camera is not the same thing as understanding.
This distinction matters because oral assessment is not automatically valid or reliable. A 2024 systematic review by Nallaya and colleagues screened 2,657 records and included 17 studies. Its central conclusion was conditional: the validity, reliability, and capacity of oral assessment to reduce academic-integrity breaches depend on sound design, scaffolding, and implementation.
Oral work should therefore be one part of an assessment strategy, not a universal replacement for written work or secure assessment. The University of Sydney's two-lane approach offers a useful distinction: appropriately designed open assessments can allow relevant tools, including AI, while secure in-person assessment can assure independent performance where that is necessary.
An asynchronous recording belongs on the open side of that distinction. It can strengthen the evidence available to an instructor, but it is not a secure exam.
Five useful asynchronous oral-assessment patterns
The strongest prompt is usually narrower than “summarize your work.” It asks the student to perform one intellectual action.
1. Defend one decision
Use this with an essay, project, design, lab report, or case analysis.
What did you choose, what alternative did you consider, and what evidence led to your decision?
The response should point back to the submitted work without retelling the whole thing. For a complete prompt, rubric, and setup sequence, use the 90-second oral-defense guide.
2. Explain one concept
Use this when students need to make relationships or processes clear.
Explain this concept to someone who understands the prerequisite but has not taken this course.
Set a boundary around the concept. A short teach-back is more revealing when the student must select the important relationship instead of reciting a chapter.
3. Critique evidence or a counterexample
Use this for source evaluation, disciplinary judgment, and argumentation.
Which piece of evidence most weakens this claim, and why?
Give students enough context and access to the relevant material. The goal is reasoning, not memory under pressure.
4. Reflect on a revision
Use this after feedback, drafting, studio critique, or portfolio work.
Show one change you made, explain why you made it, and identify one issue that remains unresolved.
Requiring a specific change keeps the response connected to the work rather than producing a generic reflection about effort.
5. Demonstrate and transfer
Use this for procedures, problem-solving, design walkthroughs, and applied learning.
Demonstrate your approach, then explain how it would change under this new condition.
Assess the process and explanation separately from camera or editing skill unless media production is itself a learning outcome.
Choose the medium from the evidence you need
Asynchronous video is not the best answer to every assessment problem.
Use written work when the outcome is disciplinary writing, structure, or precise written argument. Use a live oral assessment when immediate, adaptive questioning is essential. Use secure assessment when students must demonstrate independent performance under controlled conditions.
Asynchronous oral assessment fits when students need to explain, apply, critique, reflect, or demonstrate, but simultaneous presence is unnecessary. It can become an oral discussion when peer interpretation and response are part of the learning goal.
If you are still choosing among text, audio, video, and live formats, compare higher-education discussion-board alternatives before choosing a tool. If video is already the right medium, compare student video-response workflows by access, moderation, feedback, and institutional fit.
Make six design decisions before choosing a tool
Start with the assessment rather than the recorder:
- Evidence: What should the student make observable—an explanation, application, critique, or judgment?
- Anchor: Which artifact, source, problem, or experience will ground the response?
- Prompt: Should the student explain, defend, critique, revise, or transfer?
- Boundaries: How long is the response, what resources are allowed, and may students retry?
- Audience: Is the response instructor-only, private to the course, moderated before sharing, or visible to peers for a reason?
- Response: Which criteria will be scored, and when would a follow-up add meaningful evidence?
These decisions prevent a common failure mode: moving a generic discussion question into video and making the same activity harder to produce and review. The medium should contribute something the learning outcome needs.
Keep the workload bounded
Asynchronous delivery removes appointment scheduling. It does not remove the time required to review student work.
Keep informal explanations short, use a small rubric, and do not promise every student an individual follow-up. A second turn should be triggered by a useful reason: a claim needs clarification, a peer raises a substantive challenge, or the student can extend the reasoning.
For a larger class, small groups can work like asynchronous breakout rooms. Place students in stable groups around the same prompt or assign each group a different case. A group can discuss the task, divide roles, and publish one shared explanation, defense, or presentation instead of one video per student. Pair groups for focused peer critique so students respond within a manageable circle rather than reviewing the whole class.
Rotate roles such as facilitator, evidence checker, presenter, and respondent across activities. Stagger due dates, sample submissions for instructor feedback, or ask groups to nominate the moment where they most need a response. If the learning outcome requires evidence from each student, pair the group video with a brief individual reflection, contribution note, or targeted follow-up. A shared submission can assess collaboration and collective reasoning, but it should not stand in for individual evidence when individual mastery is the goal.
There is evidence that asynchronous student-generated video can operate at large enrollment. Guffey and colleagues studied an introductory physiology course across four semesters and 1,100 students. Students created asynchronous videos explaining major physiological concepts, and the researchers reported positive associations between the activity and relevant examination performance.
That is a useful feasibility example, not a universal workload promise. It comes from one course and one implementation. Pilot the format on one low-stakes activity, observe how long review takes, and adjust before making it weekly or high stakes.
Reduce anxiety and design for access
Students may be unfamiliar with oral assessment, and recording can introduce stress that has little to do with the learning outcome. Cornell recommends explaining the format, sharing criteria, providing practice, and giving students ways to prepare.
For a short asynchronous activity:
- provide the prompt and rubric in advance
- allow notes and reasonable retries unless spontaneity is being assessed
- do not grade eye contact, accent, camera confidence, background, or polish
- allow audio, screen sharing, or another mode when a visible face is unnecessary
- plan for captions, transcripts, device access, bandwidth, and privacy
- use the institution's accommodation process to provide an equivalent route
- avoid making public performance the default
The objective is not to make every student perform the same relationship with a camera. It is to give each student a clear and supported way to demonstrate the intended learning.
Academic integrity without surveillance
Tell students what AI use is permitted for the assignment: prohibited, permitted with disclosure, or expected for a defined step. Then align the oral prompt with the learning that still needs to be demonstrated.
Do not describe the result as “AI-proof.” If a response raises a concern, follow the institution's normal process rather than treating hesitation, accent, or presentation style as evidence of misconduct.
The constructive promise is smaller and more defensible: a written product shows one dimension of the work; a focused oral response can make another dimension visible.
Frequently Asked Questions
What is an asynchronous oral assessment?
It is a spoken response that a student records on their own schedule. The response is usually anchored to a prompt, problem, or piece of work and asks the student to explain, defend, critique, reflect, or demonstrate. Unlike a live oral exam, the student and instructor do not need to be present at the same time.
Can asynchronous oral assessment prevent AI-assisted cheating?
No. A recording cannot prove that a student worked without AI or completed every part of an assignment independently. It can add evidence of learning by asking the student to explain a decision, apply a concept, connect a claim to evidence, or respond to a focused follow-up. If independent performance must be assured, use an appropriately designed secure assessment.
How long should an asynchronous oral response be?
Use the shortest limit that gives students enough time to demonstrate the intended learning. A focused explanation or defense may need only 60 to 120 seconds, while a demonstration may need longer. Review time grows with every additional minute, so pilot the prompt and confirm that a strong response can fit within the boundary.
Should students have to appear on camera?
Only when seeing the student or their process contributes evidence required by the learning outcome. Otherwise, audio, screen sharing, slides, or another accessible mode may be enough. Do not grade eye contact, accent, background, camera confidence, or production quality unless those skills are explicitly part of the assessment.
How can instructors grade oral responses consistently?
Use a short rubric tied to observable reasoning rather than presentation polish. Criteria might include the accuracy of the explanation, use of evidence, quality of judgment, or ability to address the prompt. Share the rubric and an example in advance, provide a low-stakes practice attempt, and use a focused follow-up only when it would add meaningful evidence.
Can asynchronous oral assessment work in a large class?
Yes, if the workload is deliberately bounded. Keep responses short, score only a few criteria, sample or stagger activities, and avoid promising an individual follow-up for every submission. Instructors can also divide the class into small, stable groups that work like asynchronous breakout rooms. Each group discusses the same prompt or an assigned case, publishes one shared video, and exchanges a focused critique with a partner group.
Rotate group roles so participation does not always fall to the same student. When individual understanding must be assessed, add a brief individual reflection, contribution note, or targeted follow-up rather than relying on the group video alone. Start with one low-stakes activity and measure the review time before using the format more often or for higher-stakes work.
From recording to a purposeful second turn
Many tools can collect a video file. The more interesting assessment workflow begins when an explanation receives a useful response.
In Vivipod, an instructor can create a focused Topic, students can publish
short Posts, and the exchange can continue through comments, @mentions, or
timestamped feedback when comments are enabled. The second turn is optional:
ask a targeted question, invite a peer critique, or point to the moment where a
claim needs clarification.
Comments and notifications support that return to the conversation, while Canvas and web embeds let the LMS remain the course home.
Start with one pattern
Do not redesign an entire assessment system at once. Choose one place where the current artifact hides reasoning you genuinely need to assess. Pick one pattern, set a short boundary, share the criteria, and run it as a low-stakes pilot.
If written work is the anchor, start with the copy-ready 90-second defense. Otherwise, create a free Vivipod Space, choose one focused prompt, and hear what another mode of evidence adds.