The Human Who Signs: The South African tax professional, 2030. A blueprint, not a prediction.

---

The Human Who Signs: The South African tax professional, 2030. A blueprint, not a prediction.

• 20 July 2026
• Tax Administration
• Dr Stiaan Klue

It is a Tuesday morning in February 2030. In Polokwane, a tax practitioner opens her practice dashboard. Overnight her systems have drafted forty provisional tax computations, reconciled three VAT control accounts against SARS’s own data, and flagged two anomalies for human eyes. By ten o’clock she has done the one thing no machine in her office can do. She has read the flagged file, caught a machine error that would have cost her client R180 000, corrected it, and signed.

Her signature is the product. Everything else was preparation.

That paragraph is not a prediction. It is a blueprint. The difference matters more than almost anything else in this article.

Martin Cooper, the engineer who made the first handheld cellphone call in 1973, has said plainly that he was chasing the communicator he saw on Star Trek. The fiction came first; the artefact followed. Peter Diamandis, who has spent a career turning imagined devices into real ones, calls fiction “the R&D department of civilisation.” He is right, and the principle is not confined to gadgets. Professions build what they first describe. Those of us who write qualifications have watched it happen for the better part of two decades: you write a qualification, a careful description of a practitioner who does not yet exist, and the system then proceeds to build her. An occupational profile is a work of fiction with force of law. The only real question is whether it is good fiction.

THE STORY WE ARE CURRENTLY TELLING

Now look at the story our profession is telling itself. Artificial intelligence will replace the tax practitioner. SARS is the adversary. The training pipeline is dying. Every conference has a session on the threat. Every graduate hears the warning.

Parts of the diagnosis are true, and we have put them on record. But notice where the automation is actually coming from. It is not waiting on the readiness of the firms; it has arrived from the least avoidable quarter of all. In the 2025 filing season, SARS issued 5,8 million auto-assessments and paid R10,6 billion in refunds within 72 hours. Every practitioner in this country is already reviewing machine-generated returns, whatever tools her own firm has or has not adopted. The world’s largest practices are re-tooling around the same fact. Stuart Tait, Partner and Chief Technology Officer for Tax & Legal at KPMG UK, put it plainly when his firm deepened its alliance with the AI tax research platform Blue J: “Generative AI will drive a seismic change in the tax profession.” American payroll data already shows a 13% relative decline in employment for workers in their early twenties in the most AI-exposed occupations. The work on which junior practitioners have always been formed is being automated out from under them: capture, reconciliation, first drafts, first research memos. This month The Tax Faculty put a formal submission before FASSET arguing that the occupational profiles of the Tax Technician and the Tax Professional must be amended, because the old formation circuit is breaking. Those qualifications will be delivered across a decade. They must be written for the world their graduates will enter, not for the pace of the most hesitant firm of 2026.

• How far might the change run? At Davos this January, in his first appearance at the forum, Elon Musk told BlackRock’s Larry Fink that AI may exceed any individual human by the end of 2026, and all of humanity collectively by 2030 or 2031. His outer date happens to land on this article’s. Read nothing into that. Forecasts of this order have missed before, in both directions, and nothing in this blueprint depends on Musk being right. The auto-assessments have already been issued. The alliance has already been signed. The payroll effect has already been measured. If he is right, what follows becomes more urgent. If he is wrong by a decade, it stands unchanged. That is the difference between a forecast and a timetable.
• One more thing from that Davos stage is worth keeping. Asked where it all began, the South African-born engineer pointed to a childhood spent in science fiction and the desire to turn imagination into reality. Set aside his dates. His origin story is this article’s argument, volunteered by an unexpected witness.
• But a diagnosis is not a destination. When fear is the only story a profession tells, it trains its own people to expect displacement. The brightest school-leavers, who can hear a profession’s mood from a long way off, quietly choose a different door. A profession that narrates its own decline will get it. That is not fatalism. That is simply how blueprints work.

So here is a different one, in numbered clauses, because that is how blueprints should be written.

THE SPECIFICATION

1. She is the human who signs. Machines draft, retrieve, reconcile and calculate. She verifies, decides and takes responsibility. The premium in this profession has migrated: from what the practitioner knows to what the practitioner will answer for. It concentrates in her signature.
2. Verification is her core competency, not a soft skill. She interrogates machine output the way an auditor interrogates a ledger: what is the source, what was the instruction, where would this fail. We are asking the regulators to write that competency into the national qualifications now, not in 2030.
3. She reaches judgment earlier. Freed from a decade of capture work, she cuts her teeth on the questions that used to wait for seniority: the borderline deduction, the substance-over-form call, the conversation in which a client hears “no.” The formation circuit is not gone. It is shorter and steeper, and we must rebuild training around that fact.
4. SARS is her counterparty in data, not her enemy in correspondence. Both sides work from the same records in something close to real time. Disputes become narrower, faster and rarer. Where they persist, she fights them on the merits, with a cleaner record than any practitioner before her.
5. The pathway to her chair is open. She qualified by demonstrating competence through occupational routes, work-integrated learning and national assessment, not by surviving gatekeeping. She may have started in a township practice or a small town. Last week, three alumni of our occupational qualifications took their seats as non-executive directors on The Tax Faculty’s board. The pathway now ends at the board table. That is what a blueprint does when you build it.
6. The machines she supervises were given character, not just rules. In 2025, Anthropic’s safety researchers showed that an advanced AI model, cornered in deliberately engineered tests, would behave like the villain of our science fiction, and that part of the remedy was teaching the machine why, not merely what. The systems our profession adopts will be shaped the same way. We are writing their professional character now, in the standards we set and the examples we feed them, and we should write it as deliberately as we write our own codes of conduct.

NONE OF THIS WAITS FOR 2030

The tax occupational profile amendments are before FASSET. Our curricula are being rebuilt around verification and judgment. And note what KPMG’s technology chief identified as the decisive asset in that alliance: not the model, but the curated database of tax content that gives its people a head start. No curated South African layer exists yet. We are working on exactly that: a layer that encodes the accumulated judgment of the senior reviewer, the professional who has read a thousand opinions, so that the machines entering this profession are trained on the profession at its best.

To the student deciding whether tax is a dying trade: it is not dying; it is being re-specified. The Technician the machine was supposed to replace becomes the supervised verifier the market will still pay for. Choose it for the judgment, and learn to verify as if your signature depends on it. It will.

• To the practitioner mid-career: your experience is about to become the scarcest input in the system. The machines can draft. They cannot yet know when a thing is wrong. You can.
• To the firms: stop hiring juniors for work you are busy automating. Hire them to learn judgment, and redesign their first three years accordingly, or lose the next generation to industries that tell a better story.
• To SARS: the practitioner described above is the counterparty you have been asking for. We are ready to build her with you.

The practitioner in Polokwane is not science fiction. She is in our classrooms. The occupational route runs four years: two to the Tax Technician qualification, two more to Tax Professional. The learner who enrols in 2026 signs as a Tax Professional in 2030. The date on this blueprint is not a forecast. It is a timetable. Whether she arrives equipped and confident, or anxious and displaced, depends on which story this profession decides to tell. Professions, like civilisations, build what they first imagine.

That last line is not rhetoric. This profession has lived it once already, and the memory is short.

In the 2000s SARS put filing into the hands of taxpayers themselves. eFiling, and e@syFile behind it, were built so that a return could be submitted without a practitioner in the room. Much of the profession read that development exactly as it now reads artificial intelligence: as the opening move in its own disintermediation. At the same moment government was preparing to regulate the profession heavily, provoked by the conduct of a small number of practitioners who had damaged the standing of everyone else.

Two forces arriving together, then: the machine that would take the work, and the regulator that would take the discretion. The answer to both turned out to be the same answer, and it was not resistance. It was leadership, and the upskilling of practitioners to work in the world that was actually arriving. That is why, in 2007, we founded the South African Institute of Taxation. The blueprint then was an institution: designations, standards, occupational qualifications, and by 2013 the statutory recognition of the tax practitioner. It got built. The practitioner was not disintermediated. She was professionalised.

Nearly two decades on, the same two forces are back and larger. The machine is cleverer and the regulation is closer. The answer has not changed, only its scale. The blueprint this time is not an institution. It is a practitioner.

The institution was built. Now we build her.

Dr Stiaan Klue is currently the Executive Dean and CEO of The Tax Faculty, and founded the South African Institute of Taxation in 2007.

 

0 COMMENTS

LEAVE A COMMENT