If we don't harness technology effectively to make some elements of teaching, learning, assessment and feedback more efficient, how much more could it cost us - in academic time and contractors - to satisfactorily prove our students learned what our credential says they did?

By Gene Blackley, Co-founder & Head of Learning Technology, Assurar

For independent higher education providers, that is a question for the executive team and the board, not a matter to be delegated down to an academic committee. It sits at the exact point where two of the hardest pressures on independent providers now converge: cost, and credibility. Generative AI has dramatically shifted that point, and most providers are still looking at only one half of the equation.

A hidden cost line?

Independent providers operate without the cushions a public university leans on. No Commonwealth grants, no research income, smaller cohorts, less buffer to absorb a bad year. Every cost expansion is felt more directly, which can make the costs that aren't clearly itemised the dangerous ones.

There is one such cost set to grow significantly across the sector right now, and it rarely appears as a tidy line item. Call it the cost of proving that learning is genuine - the cost of assurance of learning. Since AI became capable of producing a highly competent essay, report or other take-home submission, the old assumption - that a piece of submitted work reflects the student's own thinking - no longer holds.

Re-establishing that confidence takes effort, and the effort is likely to be smeared and hidden across the budget, or taking a hidden toll on academics. Some of it is redesign - academic time spent rebuilding programmatic assessment or designing new assessment tasks that are harder to game. Some of it is an expansion of supervised, in-person assessment, with the invigilation, scheduling and venue costs that come with it. The rest surfaces as more sessional markers and external moderators, and potentially hours lost to misconduct investigations and academic appeals. Few organisations total these costs, so few organisations are actively managing it - but it is real, and it is rising. Or it’s being absorbed by individual academics as longer hours each week that will be quietly eroding teacher satisfaction, energy levels and quality - building to burn out and attrition.

For an independent provider, the exposure goes well beyond the budget and margins. Assessment quietly underpins three key differentiators that genuinely set the college apart. The quality and reputation of the credential rests on whether its results can be trusted. Student experience and satisfaction are shaped directly by the accuracy, fairness, quality and timeliness of grading and feedback. And the employability links to industry depend on being able to stand behind a graduate and say, with confidence (and evidence), that they genuinely understood what they were taught. Weaken the integrity of the assessment underneath, and all three soften at once.

In brief:

The cost of proving students - not AI - did the work is rising.
Assuring learning underpins an independent provider’s three key differentiators: credential reputation, student experience, employability.
AI cuts both ways - teachers can harness it to elevate quality of assessment & feedback.
Interactive oral assessment best proves real understanding; now, real-time AI orals can scale it.
The prize: AI assesses student knowledge and drafts grading and feedback at scale. Educators refine and approve it - returning academic time.

AI: a tool that cuts both ways

It is tempting to treat AI in education as either a threat to be policed or a miracle to be adopted. It is neither. Almost any technology amplifies whatever application and judgment sits behind it. Pointed well, AI can lift the quality and consistency of teaching, learning, assessment and feedback. Pointed carelessly, it can hollow them out - automating away the very thinking and effort that is supposed to build the learning and knowledge.

The useful question is therefore not whether to use AI, but where it elevates learning and where it degrades it. Where it can enhance equity and accessibility, versus where it can exacerbate inequality.

AI is incredibly valuable for automating assessment and grading, drafting truly personalised feedback, and for surfacing patterns across a cohort that a tired marker may miss. It is corrosive when it becomes a substitute for the student's own reasoning, or when it is used to mass-produce assessment judgments with no human-in-the-loop to assess and refine the final outcome published to the gradebook and the student. The same tool, two outcomes - separated entirely by how it is applied and who remains in command of it.

The prize providers could miss out on: academic time

Here is the half of the equation that gets overlooked. Marking and individual feedback are among the most labour intensive, least enjoyable and most inconsistent parts of teaching. We know they are inconsistent because we built an entire apparatus - moderation - to correct for the bias, fatigue and drift that creep in when a human marks the fortieth submission differently from the fourth. That apparatus is itself a cost, and a tacit admission that the process is imperfect.

This is precisely where AI earns its place. It can draft assessment and feedback across an entire cohort to a consistent standard, with educators directing, reviewing, refining and finalising every output before it reaches a student or a gradebook. The educator's judgment stays decisive; what changes is that they begin from a complete, consistent first pass rather than a blank page repeated forty times, or a template applied lazily without sufficient nuanced customisation to each submission.

The effect is twofold. It lifts the floor - every student receives feedback of a consistency and depth that is hard for humans to sustain manually by the end of a marking run, which is itself an equity gain. And it returns academic time: the scarcest, most expensive resource an independent provider has.

That returned time is not an efficiency to be banked by increasing student numbers or adding more subjects to a teacher’s load. It is time that teachers can redirect to the highest-value work technology cannot do for them - the direct, personal interaction with students that powerfully moves learning forward. The goal is not to remove educators from assessment. It is to remove them from the arduous, repetitive toil so they can spend more of their week on the joyful and fulfilling elements of being an educator.

It is worth addressing a concern that has grown louder lately: students have begun to object to AI being used to mark their work, often seeing it as unfair or impersonal - or as a double standard when they themselves are told not to use it. That objection largely dissolves once the arrangement is explained honestly: the teacher remains the judge, the AI only does the first pass, and the student is the one who benefits: fairer and more consistent grading, more thorough personalised feedback delivered sooner, and more of their teacher's time to discuss it directly.

The verification gap

Reclaiming time solves the cost half. It does not, on its own, solve credibility - and this is where the two halves rejoin. If we can no longer assume an unsupervised submission reflects a student's own thinking, we need a way to confirm comprehension that AI cannot complete on the student's behalf.

Educators near-universally agree on what that method is. An interactive oral assessment - a structured conversation that probes understanding in real time, with follow-up questions a student cannot anticipate or delegate - is the most thorough way to evaluate genuine comprehension across almost any unit of learning.

Oral assessment has been considered a gold standard, almost foolproof assessment method for centuries. Its problem has never been effectiveness. Its problem has always been scale: one-to-one oral examination is slow, costly to staff, and hard to run consistently, so while many higher-ed institutions are now increasingly discussing how they can bring it back and use it more, many are talking about using it once per semester at best. But that is too infrequent to be fair or reliable for diverse student cohorts (think extroverts versus introverts), and too dependent on which examiner a student happens to draw.

So the limiting factor for using more interactive oral assessment is the vast cost of delivering it at scale. If we can remove that constraint, what becomes possible for improving and accelerating student learning, and addressing academic integrity?

One move, two problems solved

The opportunity is that a single, deliberate shift addresses both halves at once. AI-assisted assessment and feedback, drafted at cohort scale under educator direction, returns academic time, reduces bias and raises consistency. Scalable, live, real-time interactive oral assessment with AI, with educators guiding the design and approving every result, restores confidence that the learning behind the credential is real. The same intervention relieves the cost pressure and the credibility pressure together - higher quality, cheaper, and less risk.

That kind of scaled, educator-controlled, AI interactive oral assessment is now available rather than aspirational. It is the problem we work on at Assurar - but the strategic choice is the same whatever tools you use: keep absorbing the rising cost of assuring learning, or move deliberately to get ahead of it.

The board-level reframe

For an independent provider, defending the credibility of the credential and easing the academic cost squeeze are not two separate projects competing for the same budget. Handled well, they are the same project. The regulators' direction of travel - toward providers being able to demonstrate that their assessment still proves what their qualifications claim - only sharpens the point.

The cost of proving learning in the AI era is already being paid, quietly, and increasingly, in academic time and contractors. The only real decision is whether to keep paying it by default, or to harness the technology deliberately and turn an unmanaged cost into a genuine source of advantage. The providers who get assessment right will not just protect their position. They will widen the gap on the ones who continue to treat it as a compliance chore.

For independent providers, what’s the cost to prove learning in the AI era?