AI changes how KPMG accountants build judgment and skills

The missing apprenticeship

AI is taking over the repetitive, low-risk work that used to serve as the apprenticeship for young accountants, and that changes the bargain inside a firm like KPMG. The old model was simple: do enough routine testing, reconciliations, and control work, and judgment, skepticism, and systems knowledge would come with repetition. That path is disappearing, which means the profession has to replace accidental learning with deliberate training.

That is the core warning in the Journal of Accountancy’s March feature: if automation absorbs the work that once taught systems, controls, and professional skepticism, firms cannot assume those skills will develop on their own. For KPMG staff, that is not an abstract talent debate. It reaches directly into promotion cycles, the partner track, and the way first and second year work gets used to prove readiness for more responsibility. If the entry-level ladder loses its lowest rungs, the next question is blunt: what exactly are people expected to learn instead, and who pays for that learning?

What juniors used to get from repetitive work

In audit and advisory, the first years often looked like a long apprenticeship in doing the same kind of work over and over. That repetition was not glamorous, but it was how people learned how systems behave, how controls fail, and how to spot when an answer is technically correct but professionally thin. It also created the habit of skepticism, which is hard to teach in a classroom and easy to take for granted when there is always another test, another tie-out, another review note.

When AI handles more of that baseline work, the learning curve gets steeper. A junior who skips too much rote work may move faster on paper but reach review meetings with less context for why a result matters, what a control is actually protecting, or how to challenge an automated output. In busy season, that can turn into a real career risk: fewer safe repetitions, less room to learn by doing, and more pressure to sound ready long before the person has had enough reps to be ready.

The new bargain: more intentional learning

The answer the profession is circling is not less training, but better targeted training. The Journal of Accountancy piece points toward conceptual mastery, human judgment, technological fluency, and adaptability as the capabilities that matter more when machines do the routine work. In practice, that means more simulations, case-based exercises, stretch assignments, and structured coaching, rather than a long run of identical entry-level tasks.

For KPMG professionals, that shift matters because it changes where development happens. Learning will increasingly come from review conversations, client-facing problem solving, and assignments that force people to explain why an answer makes sense, not just whether the template is filled out correctly. That is a higher-value path if it is done well. It is also more demanding, because it asks managers to create judgment-building moments earlier and more deliberately, instead of assuming the work itself will teach them.

How KPMG is building the platform underneath it

KPMG has already staked out a version of this future. On April 23, 2025, KPMG LLP said it was accelerating AI integration into KPMG Clara, its global smart audit platform, and said AI agents would be deployed to empower more than 95,000 auditors globally while keeping a human in the loop. The promise there is efficiency, but the real operational change is that humans are supposed to spend less time on the mechanical parts of audit and more time on the calls that shape quality, insight, and trust.

KPMG International went a step further on June 17, 2025, when it launched KPMG Workbench as a foundational single AI platform to scale global adoption and integration of AI. That kind of platform matters because it is not just a tool rollout. It is an attempt to standardize how the firm uses AI across geographies and teams, while keeping control and trust at the center. KPMG’s audit and technology messaging says the same thing in different language: the firm is transforming the audit experience through process automation and enhanced agility so professionals can focus on the data that matters most and offer deeper insights.

For staff, the practical effect is straightforward. If AI is absorbing some of the rote work, KPMG has to make room for earlier exposure to analysis, review, and client discussion. Otherwise, the firm risks building a workforce that is fluent in prompts and platforms but thin on judgment. That is a bad trade in audit, and it is a worse one in advisory, where credibility depends on more than speed.

Why this is also a control problem

The regulatory backdrop makes the training question more serious. On December 20, 2024, the SEC announced that it had approved two PCAOB proposals updating audit standards, including standards related to technology-assisted analysis in an audit. That alone signals that the profession is moving into a period where the rules around digital work are still being clarified, not fully settled.

Then, in a September 16, 2025 speech, PCAOB Board member Christina Ho said there are still open questions about acceptable AI-based audit approaches and how compliance should be assessed. That is the part firms cannot ignore. Training is not just a retention strategy or a perk for high-potential staff. It is a risk-control system. If people do not know how to evaluate automated outputs, challenge them, and document the reasoning behind a conclusion, audit quality and client trust are both exposed.

What the profession is doing now

KPMG is not the only firm trying to solve this. The AICPA and CIMA have launched AI skills programming for accounting and finance professionals, a sign that the broader profession is treating AI fluency, governance, and judgment as core competencies rather than optional extras. The 2025 CPA.com AI in Accounting Report, built from the AICPA and CPA.com AI in Accounting and Finance Symposium, ecosystem research, and practitioner interviews, points in the same direction: AI is no longer a side project for technical specialists. It is becoming part of the operating model.

That leaves KPMG workers with a clear, if uncomfortable, reality. The future path to leadership will depend less on surviving endless repetition and more on proving judgment earlier, in more visible ways. The firm can save time with AI, but it cannot save itself from the cost of teaching people how to think.