Editor Note

Every professional now uses AI. Very few have been taught when to trust it.

This week’s guest edition comes from Dr. David Z. Jacome, an educator, TEDx speaker and valued member of the MBA Leaders Forum board who consults with accounting firms on AI governance. Building on Priya Udeshi’s recent edition, David takes the governance conversation into the classroom, where tomorrow’s leaders are learning to use AI long before anyone teaches them to judge it. Whether you lead a team, hire graduates, or are building your own skills for the next stage of your career, his argument is one you’ll want to read closely. Thank you, David, for this thoughtful and timely contribution.

Executive Summary

Priya Udeshi made the case in a recent edition of this newsletter that the future of work is not about which jobs AI will replace. It is about whether organizations can govern what comes next. She is right, and her argument deserves to be extended into the two domains where I spend my professional life: the accounting profession and the business school classroom.

The accounting and finance function has adopted AI faster than almost any corner of the enterprise. The Consero Global 2026 CFO Report found that 97 percent of finance departments have now adopted AI in some form, up from 76 percent just a year earlier [1]. Yet Thomson Reuters’ 2026 survey of more than 1,500 professionals found that only 18 percent of organizations track AI return on investment at all [2]. Adoption without measurement. Experimentation without evaluation. That is the same governance gap Priya described, playing out in the profession whose entire identity is built on measurement, controls, and assurance.

The deeper problem is upstream. The professionals who will be expected to close this gap are sitting in business school classrooms right now, and we are largely not teaching them how. Surveys consistently show that between 80 and 92 percent of students already use generative AI in their studies [3, 4]. Meanwhile, an AACSB survey of deans and faculty found that nearly 45 percent of business schools lack the coordinated governance needed to translate their own AI policies into practice [5]. We are producing graduates who are fluent in using AI tools and unprepared to govern them. That is not a skills gap. It is a judgment gap, and it is the one that will define the next decade of the profession.

The Profession That Measures Everything Is Not Measuring This

Accounting is the discipline of verification. We built our professional identity on internal controls, audit trails, materiality thresholds, and the principle that a number you cannot support is a number you cannot report. So it should trouble all of us that the profession’s own AI adoption looks, from a governance standpoint, remarkably undisciplined.

Consider the pattern in the current data:

Adoption is nearly universal, and it happened fast. Beyond the 97 percent finance department figure, Thomson Reuters found that organizational use of generative AI among tax and accounting professionals jumped from 22 percent to 40 percent in a single year, and that 86 percent of tax professionals using generative AI do so at least weekly [2, 6]. Intuit’s 2026 survey of 725 U.S. accounting professionals found 88 percent using AI for client services and 86 percent for firm operations [7].

Measurement has not kept pace. Only 18 percent of organizations in the Thomson Reuters survey track AI return on investment [2]. A CFO.com survey of 321 U.S. finance and accounting decision-makers found that only 28 percent of finance teams using AI are seeing measurable financial impact [1]. The firms that do see returns look structurally different: Grant Thornton’s 2026 CFO research found that fully integrated firms are four times more likely to report revenue growth than pilot-stage organizations [1]. Integration, not experimentation, is what pays. And integration is a governance outcome.

Risk controls lag adoption. KPMG research found that 46 percent of U.S. accounting firms have inadvertently input confidential client information into public AI services [6]. Read that again. Nearly half of the firms in a profession bound by confidentiality obligations have already had a data governance failure with AI, most without a policy framework in place to prevent the next one. Thomson Reuters also found that 61 percent of firms have stalled or halted AI projects due to skill shortages [6].

None of this is a technology failure. The tools work. What is missing is exactly what Priya identified at the enterprise level: the decision infrastructure. Which AI use cases align with the firm’s strategy and risk appetite? What does value mean for each one? Who decides what client data can enter which system? What triggers a shutdown? How is return measured, and who is accountable for it? These are portfolio governance questions, and in most firms I work with, nobody owns them.

I see this firsthand in my consulting practice with accounting firms. The typical engagement does not begin with “help us choose an AI tool.” It begins with a firm that has already accumulated a dozen tools, three unofficial ChatGPT workflows the partners do not know about, and no shared answer to the question of what any of it is for. The first deliverable is never technology. It is an inventory, a data classification policy, an evaluation framework, and a decision-rights map. Governance first. Tools second. The firms that reverse that order become part of the statistics above.

The Second Governance Gap: The One in the Classroom

Here is where I want to push the conversation somewhere Priya’s piece pointed but did not go. Every governance framework, every portfolio process, every measurement system she called for has to be designed, operated, and defended by people. Where do those people learn to do it?

Increasingly, the honest answer is: nowhere.

The student side of the equation has moved with breathtaking speed. The UK’s Higher Education Policy Institute found that the share of students using generative AI tools jumped from 66 percent to 92 percent in a single year [3]. Stanford’s 2026 AI Index estimates that four in five university students now use generative AI, and Chegg’s survey of nearly 12,000 undergraduates across 15 countries found that 80 percent use it to support their studies, with a majority of U.S. users prompting at least once a day [4]. Our students are not waiting for permission or curriculum. They are already fluent operators.

The institutional side has moved far more slowly, and unevenly. To be fair, there is real progress. The AACSB, GMAC, and Graduate Business Curriculum Roundtable framework for AI in business education has grown from 26 participating schools in July 2025 to 84 in July 2026, and it documents a genuine shift from experimentation toward institution-wide integration [8]. AACSB’s proposed 2026 Global Standards elevate digital agility and AI literacy toward accreditation expectations rather than optional innovation [9]. These are the right moves.

But look closely at what the same body’s research reveals: nearly 45 percent of business schools lack the coordinated governance to translate their own generative AI policies into practice [5]. Business schools, in other words, have the identical failure mode as the organizations they study. Scattered faculty innovation, inconsistent policy, no shared definition of what AI competence means, and limited mechanisms for scaling what works and retiring what does not. The AACSB’s own contributors have proposed six-pillar governance frameworks for schools precisely because the ad hoc approach is producing innovator fatigue and inconsistent institutional impact [5].

And even where AI has entered the curriculum, notice what we are teaching. The dominant model is tool fluency: prompt this system, automate that workflow, build this dashboard. Tool fluency matters, and I teach it. But tool fluency is the easy half. The World Economic Forum’s Future of Jobs Report identified skills gaps as the single biggest barrier to business transformation, cited by 63 percent of employers, and estimates that 39 percent of workers’ existing skills will be transformed or become outdated by 2030 [10]. The instinctive institutional response has been to teach more tools faster. But the skills that will not become outdated are the governance skills: knowing how to evaluate an AI investment against strategic priorities, how to define value before deployment rather than after, how to design controls around a probabilistic system, and how to decide when to scale and when to stop.

Nobody’s prompt library will matter in five years. Their judgment will.

Why Accounting Education Carries a Special Burden

This argument applies across business education, but it lands hardest in accounting, for three reasons.

First, the pipeline math leaves no slack. More than 300,000 accountants and auditors left the profession or the labor force between 2019 and 2022 [11]. The AICPA’s 2025 Trends report shows schools awarded 55,152 accounting bachelor’s and master’s degrees in the 2023 to 2024 academic year, down 6.6 percent from the prior year, extending a decade-long decline even as the Bureau of Labor Statistics projects over one hundred thousand accounting and auditing openings each year [12, 13]. There are encouraging signals: accounting enrollment rose 8.9 percent in spring 2026, the third consecutive year of growth [14]. But every graduate we produce now has to carry more responsibility, earlier, than any prior generation. A shrinking profession cannot afford graduates who need five years of on-the-job seasoning before they can be trusted with judgment. The judgment training has to start in the classroom.

Second, AI is consuming exactly the work we used to train juniors on. The traditional accounting apprenticeship ran on repetition: ticking and tying, reconciliations, workpaper preparation, first-pass research. That repetition was inefficient, but it was also where professional skepticism was quietly built. As AI absorbs that layer, the profession loses its default training ground for judgment at the very moment it needs judgment most. Some observers now argue AI is absorbing entry-level accounting work faster than the pipeline initiatives can replenish it [15]. If the bottom rung of the ladder is disappearing, education has to build the rung somewhere else, deliberately.

Third, accountants are the natural owners of AI governance, if we prepare them for it. Strip away the technology and look at what enterprise AI governance actually requires: control design, evidence evaluation, risk assessment, independence of judgment, documentation discipline, and the ability to tell leadership an uncomfortable truth with support for every number. That is not a new skill set. That is the accounting skill set, pointed at a new object. The evolved CPA exam already gestures in this direction with its Information Systems and Controls discipline [12]. The firms that see the strongest AI returns are the ones that pair tools with explicit data-handling policies and mandatory training before go-live [6]. Someone has to design those policies and controls. There is no professional community better positioned than ours, and no better recruiting pitch for the pipeline than this: the profession that governs the numbers is becoming the profession that governs the machines producing them.

What Teaching AI Judgment Actually Looks Like

It is fair to ask what I would put in the curriculum instead of, or alongside, tool training. Here is what I have been building into my own courses and what I believe the next revision of accounting and business curricula requires.

Evaluation before automation. Before students automate a workflow, require them to write the business case a governance committee would demand: the value hypothesis, the measurement plan, the risk assessment, the kill criteria. In my analytics courses, the deliverable is not the working workflow. It is the workflow plus the memo that justifies its existence and defines how we will know if it failed. Students resist this at first. It feels like paperwork. Then they interview for jobs and discover it is the scarcest skill in the room.

Verification as a first-class competency. Every AI output in my classroom arrives with a standing question attached: how would you audit this? Students learn to trace an AI-generated analysis back to source data, to identify what the system cannot know, and to document the review the way they would document audit evidence. Professional skepticism does not transfer to AI automatically. It has to be practiced against AI specifically, because probabilistic systems fail differently than careless humans do. They fail confidently.

Data governance as ethics in practice. The 46 percent confidentiality failure statistic [6] belongs in every accounting ethics course in the country, not as a scare slide but as a case study. What policy would have prevented it? What classification scheme? What training? Students should graduate having actually drafted an AI acceptable-use policy for a hypothetical firm, argued about its edge cases, and defended it.

Portfolio thinking, not project thinking. Business curricula still mostly teach technology decisions as single-project capital budgeting exercises. The reality Priya described, and the one my consulting clients live in, is a portfolio of dozens of concurrent AI initiatives competing for constrained capacity. Students need practice in comparative evaluation: given ten pilots and funding for three, which do you scale, which do you stop, and how do you defend the decision? That exercise teaches more about the future of their work than any individual tool ever will.

Cross-functional translation. The governance gap persists partly because technologists and business leaders lack a shared language for value. Accountants can be that translation layer, but only if we train them to present control and measurement concepts to non-accountants. Communication is not a soft-skills garnish on this curriculum. It is the delivery mechanism for everything else.

None of this requires abandoning tool instruction. It requires reframing it. The tool is the object of study. Governance is the discipline being taught.

The Institutional Mirror

One more uncomfortable observation, offered in the spirit of the honest debate this board has become known for. Business schools cannot credibly teach AI governance while practicing AI anarchy.

When nearly half of schools lack coordinated governance for their own AI adoption [5], the hidden curriculum undermines the formal one. Students notice when one professor bans AI, the next requires it, and the institution’s policy is a paragraph nobody can locate. What they learn from that experience is that governance is theater. The schools documented in the AACSB framework as exemplars share a common trait: they treated their own institution as the first case study, pairing platforms with governance frameworks, policies, and oversight before scaling into the curriculum [8]. That sequencing is the lesson. The 4P framing some AACSB contributors propose, covering people, policy, pedagogy, and platform, is a reasonable starting scaffold for any school still improvising [16].

This is also, frankly, a leadership opportunity for those of us in accounting departments. We teach controls for a living. We should be volunteering to help design our institutions’ AI governance, not waiting for a task force memo. I have found that the credibility this builds flows directly back into the classroom and into relationships with the firms that hire our students.

Where This Leaves Us

Priya ended her piece by reframing the future of work question: not which jobs AI will take, but who will build the organizational systems that determine whether AI creates value or noise. I want to extend her reframe one step further upstream.

Who will teach the builders?

The governance infrastructure the next decade requires will be designed by people who today are undergraduates using ChatGPT to summarize their textbook. Whether they become the professionals who close the governance gap or the ones who widen it depends substantially on what happens in our classrooms over the next few years. The data says the profession is adopting faster than it is governing, the pipeline is too thin to waste a single graduate on obsolete training, and the institutions doing the training are struggling with the same governance deficit they need to teach against.

That is not a counsel of despair. It is a to-do list, and an unusually clear one. Teach evaluation before automation. Teach verification as rigorously as we teach preparation. Practice institutional governance so students see it modeled, not just described. And tell the next generation the truth about their profession: the ledger is being automated, but judgment is not, and judgment is now the product.

The future of work in accounting is not fewer accountants. It is accountants doing the one thing machines cannot: deciding what should be trusted, and taking responsibility for the answer.

What This Means for Your Career

You don’t have to be an accountant or a professor for David’s argument to apply to you. In nearly every field, knowing how to use AI tools is quickly becoming the baseline. What employers increasingly need, and rarely find, is someone who can judge what those tools produce.

That gives every professional three practical steps. First, check what AI gives you before you rely on it, and be able to explain how you checked. Second, when you propose using AI for a task, write down what success looks like and when you would stop. Third, keep a record of decisions you made well with AI’s help, because that record is proof of judgment, and judgment is what hiring managers are searching for.

At the MBA Standards Board™, this is why our Competency Conversation™ tests judgment, not memory. A credential should show that you can think through a real decision, not just that you completed a course.

Where in your work are you using AI without a clear way to check it? Share it in the comments. David and I will be reading.

Frequently Asked Questions

If AI adoption in accounting is already near universal, has the governance window closed?

No, and the data suggests the opposite. Adoption being universal is precisely why governance is now the differentiator. When 97 percent of finance departments have adopted AI [1] but only 18 percent of organizations measure its return [2], the competitive separation is no longer about access to tools. Grant Thornton’s finding that fully integrated firms are four times more likely to report revenue growth than pilot-stage peers [1] shows that the returns accrue to organizations that built the surrounding discipline. The window for adopting first has closed. The window for governing best is wide open.

Is this an argument against teaching AI tools in accounting programs?

Not at all. Tool fluency is necessary and students expect it; surveys show most of them arrive already using AI daily [3, 4]. The argument is about what surrounds the tool instruction. A student who can automate a reconciliation but cannot write the evaluation criteria, the data-handling policy, or the review procedure for that automation is half-trained, and it is the wrong half to stop at. The tools will be replaced repeatedly during their careers. The evaluation discipline will not.

Why should accounting faculty and firms treat this as a shared problem rather than blaming each other?

Because the failure modes are interlocking. Firms report that skill shortages have stalled 61 percent of AI projects [6], while schools report that governance deficits prevent their AI policies from reaching practice [5]. Firms need graduates trained in judgment; schools need firms to articulate what judgment looks like in current practice and to provide the case material. The most productive model I have seen is co-designed: firms bring real, anonymized AI governance dilemmas into the classroom, and faculty bring evaluation frameworks and assessment rigor. Both sides get a stronger pipeline, and the AICPA enrollment recovery now underway [14] gives us more students to build it with.

Does the shrinking entry-level layer mean the traditional accounting career path is finished?

The path is being compressed, not eliminated. The apprenticeship layer of repetitive work is shrinking as AI absorbs it, which means judgment formation has to be front-loaded into education and early career design rather than left to accumulate through repetition [15]. That is a solvable design problem. Professions from medicine to aviation deliberately train judgment through simulation and structured cases rather than waiting for volume experience. Accounting education can do the same, and the programs that figure it out first will own the placement pipeline.

From the Boardroom, Questions From This Discussion

Questions and answers drawn from the board and reader discussion of this edition, paraphrased and credited to the people who raised them.

What Is the Difference Between AI Tool Fluency and AI Judgment?

AI tool fluency is knowing how to operate an AI system, such as writing prompts or automating a workflow. AI judgment is knowing whether the output can be trusted, what it left out and when to stop using it. Dr. David Z. Jacome, who teaches accounting and business analytics, argues that the tools will be replaced several times over a career while judgment lasts. Jay Manouchehri, a CEO and board member, summed it up in the discussion. “Prompting is a commodity; discernment is the premium skill.”

What Evidence Convinces an Employer That a Candidate Can Be Trusted to Review AI Output?

The strongest evidence is a specific example with a paper trail. Dr. Jacome says the most convincing candidate can describe a time AI gave a wrong answer, how they caught it, what they checked and what they changed, and can then show the review notes. That one example proves three things. The candidate defined what a correct answer looked like before starting, tested the output with skepticism and documented the review the way an auditor would.

Can Candidates Fake a Story About Catching an AI Error?

Yes, once that story becomes the expected interview answer. James Lancaster, CMBA™, a founder and investor with 52 angel investments, warned that candidates will arrive with rehearsed stories and coaching services will sell the template. His alternative is to hand the candidate a one-page analysis with a planted flaw and ask what they would check first and why, because the order they check things in reveals more than whether they find the flaw. Dr. Jacome called this Goodhart’s law applied to hiring. When a measure becomes the target, it stops being a reliable measure.

What Interview Question Best Reveals Real Judgment?

Ask the candidate to describe a time they were wrong and caught it too late. James Lancaster, CMBA™, explained why it works. Anyone can narrate a catch, but very few people can walk through a miss honestly, and that kind of answer is close to impossible to rehearse into something flattering.

How Can Business Schools Evaluate a Student’s Judgment Before Graduation?

Grade the decision, not just the deliverable. In reply to a question from JoAnn Laing, CMBA™, a board director and digital modernization leader, Dr. Jacome described three methods. Give students AI output with a hidden error or missing risk and score whether they catch it. Ask for a decision memo explaining what they accepted, what they rejected and why. And require a short oral defense to confirm the reasoning is truly theirs, an approach he is using this semester in his econometrics course.

How Can a Professor Tell Whether a Student Understands AI-Assisted Work?

Ask the student to explain it out loud. Randy Priem, a finance professor at the University of Antwerp and a senior policy advisor at IOSCO, recommends following written work with a presentation or oral update, because a student who truly wrote something can explain it easily. Dr. Jacome put the principle in one line. “If you can’t explain it, you don’t own it.” Priem also noted that the business schools where he teaches give students a dedicated session on AI governance, covering plagiarism, critical review of output and what confidential information must never be shared with AI.

Why Is AI a Threat to How Junior Professionals Develop Judgment?

AI is absorbing the repetitive entry-level work where judgment used to be built. In accounting, that work included ticking and tying, reconciliations and first-pass research, tasks that quietly trained professional skepticism. Chandni Prasad, CMBA™, an AI product manager and startup mentor, warned that without a new entry point, organizations will face a shortage of capable mid-level leaders and, later, of senior leaders ready to be accountable for AI-led decisions. Her prescription is to front-load training in risk, ethics and validation instead of waiting for experience to accumulate.

If AI Removes the Apprenticeship, How Will Organizations Develop Experts?

Organizations will have to rebuild the apprenticeship on purpose. Josef Stetter, MBAe™, Certified MBA™, asked the question directly. If we automate the work that teaches judgment, who will govern AI when today’s experts retire? Dr. Jacome offered three answers. Pair juniors with seniors to review AI output together, capture expert judgment now while experienced people are still in the room, and reward the person who catches an error, not just the people who adopt the tools. Stetter’s advice for leaders was blunt. “Stop measuring who can use AI. Start measuring who can challenge it.”

How Many Accounting Firms Have Put Confidential Client Data Into Public AI Tools?

Research from KPMG found that 46 percent of U.S. accounting firms have inadvertently entered confidential client information into public AI services. Dominique Colvard, MPAS, MBAe™, a healthcare leader, compared it to entering a patient’s protected health information into a public chatbot, the exact risk HIPAA exists to prevent. She also drew a useful distinction. Many clinics now use AI for transcription, but through internal tools monitored by their own IT and cybersecurity teams, not public websites.

What Should You Ask Before Putting Work Data Into an AI Tool?

Ask the same questions an auditor would. Dr. Jacome teaches students to ask where the data goes, who can access it and whether it will be used to train the AI model before they use any tool. In his words, professional skepticism shouldn’t stop at the financial statements. It should extend to the tools used to analyze them.

Why Is AI’s Confidence a Risk for Decision-Makers?

AI often presents uncertain answers with full confidence, and people tend to trust confidence. Katherine Mathews, CMBA™, a medical director, recalled the hardest exam of her medical training, an oral examination by three Harvard faculty members. What they were really testing was whether students could admit what they didn’t know, recognize when there might be no right answer and tolerate uncertainty. Her conclusion was memorable. “AI’s certainty can be a fool’s gold.”

Does Relying on AI Keep People From Mastering Their Field?

It can, because knowledge in any field works like a language that has to be practiced. Katherine Mathews, CMBA™, compared relying on AI to listening to language tapes instead of speaking with a native speaker. A student who leans on AI may get the answers without ever becoming fluent in the discipline itself.

How Can Workplace Training Build Judgment and Not Just AI Skills?

Training should move people from knowing how AI works to being accountable for what they do with it. Beth Fries-Packard, DBA, CMBA™, a workforce readiness strategist, frames it as Know, Do, Become. Know AI’s capabilities, limits and biases. Do the work of verifying, questioning and challenging its output. Become a professional who uses AI with confidence while owning every decision. Dr. Jacome added that the most costly errors are not always wrong numbers. Sometimes they are a missing point of view, such as an overlooked stakeholder.

Why Does AI Judgment Matter in Sales and Commercial Leadership?

Faster analysis only creates value when someone can tell whether it fits the real situation. James Johnson II, CMBA™, a director of national accounts, said the differentiator will be less about who can use AI tools and more about who can question the output, recognize what may be missing and turn it into sound business decisions. Dr. Jacome agreed, noting that AI can summarize the data, but it doesn’t sit across the table from the customer.

What Should a Team Define Before Automating Any Workflow With AI?

Define the value hypothesis, the measurement plan and the kill criteria before anything is automated. Priya Udeshi, CMBA™, PMP, Chief of Staff to the CIO and head of the global IT program office at MongoDB, said she would like to see those three items on every project intake form, because they make the later decision to scale or stop far easier. She also confirmed that Dr. Jacome’s classroom exercise, choosing three of ten AI pilots to fund, closely mirrors how real portfolio decisions are made.

Is the Classroom the Only Place Where Judgment Is Built?

No. Sarah Ellison, Ed.D., an associate vice president of student affairs, called the ability to audit, evaluate risk and exercise professional judgment “the real enduring credential,” since the tools change every six months. Dr. Jacome added that judgment also grows through advising, leadership roles and every moment students must make a real decision and stand behind it. Sam Bodunrin, Ph.D., CMBA™, a professor and board director, agreed that the classroom is where judgment must be practiced, not just described.

How Can a Career Changer Prove Judgment When Tools Change So Quickly?

Focus on the skills that carry across every field and every tool. As Cheryl Nunn, MBAe™, noted in the discussion, AI tools change every quarter, but the ability to check an output, define success before you start and know when to stop can be built and proven anywhere. A rehearsed claim is easy to make. Showing that judgment live, on a problem with an embedded flaw, is far more convincing to an employer.

About the Author

Dr. David Z. Jacome, MBA, CMBA™, M.S., is an educator who teaches accounting and business analytics, a consultant to accounting firms on AI governance and adoption, a TEDx speaker and a member of the MBA Leaders Forum board. His work focuses on helping firms and business schools build the governance, measurement and judgment that responsible AI adoption requires.

References

[1] Consero Global, 2026 CFO Report; Grant Thornton 2026 CFO Survey; CFO.com survey of 321 U.S. finance and accounting decision-makers, 2025. Adoption, integration, and measurable-impact figures as compiled in “AI in Accounting and Finance Statistics 2026,” May 2026.

[2] Thomson Reuters, 2026 survey of more than 1,500 tax, accounting, and professional services respondents: generative AI organizational use rising from 22 percent to 40 percent year over year; agentic AI at 15 percent; 18 percent of organizations tracking AI ROI.

[3] Higher Education Policy Institute (HEPI) and Kortext, Student Generative AI Survey 2025 and 2026: student use of generative AI tools rising from 66 percent to 92 percent year over year among UK undergraduates.

[4] Stanford Institute for Human-Centered AI, 2026 AI Index Report; Chegg Global Student Survey of 11,706 undergraduates across 15 countries; Gallup surveys of U.S. college students, as summarized in “Student Use of AI in Higher Education,” USC Libraries, July 2026.

[5] Bhattacharyya, S., and Mehta, K., “Leading the AI Transition,” AACSB Insights, March 2026, citing a 2025 AACSB survey of deans and faculty finding nearly 45 percent of business schools lack coordinated governance to translate generative AI policies into practice.

[6] Thomson Reuters, “2026 AI in Professional Services Report” (weekly-use and stalled-project figures); KPMG, “AI Adoption Across Finance Functions,” 2025 (46 percent of U.S. accounting firms inadvertently inputting confidential client information into public AI services), as compiled in “AI Adoption in Accounting Firms: 2026 State Report.”

[7] Intuit / Firm of the Future, 2026 Accountant Technology Survey of 725 U.S. accounting and bookkeeping professionals, June 2026.

[8] Means, T., Inspire Higher Ed, in partnership with AACSB International, the Graduate Business Curriculum Roundtable, and GMAC, “A Framework for Artificial Intelligence in Business Education: Exemplars and Critical Themes for Successful Integration,” July 2026 update (84 participating institutions, up from 26 in July 2025).

[9] AACSB International, exposure drafts of the Global Standards for Business Education and 2026 Accounting Accreditation Standards, January 2026, including proposed digital agility expectations.

[10] World Economic Forum, Future of Jobs Report 2025: skills gaps cited by 63 percent of employers as the top barrier to transformation; 39 percent of workers’ core skills expected to change or become outdated by 2030.

[11] Ellis, L., “Why So Many Accountants Are Quitting,” The Wall Street Journal, December 2022, reporting more than 300,000 accountants and auditors leaving the field between 2019 and 2022, as discussed in The CPA Journal, February 2026.

[12] AICPA, “2025 Trends: A Report on Accounting Education, the CPA Exam, and Public Accounting Firms’ Hiring of Recent Graduates,” October 2025: 55,152 accounting degrees awarded in 2023 to 2024, down 6.6 percent; evolved CPA exam discipline structure.

[13] U.S. Bureau of Labor Statistics occupational projections for accountants and auditors, as cited in AICPA/NASBA pipeline analyses, 2025 to 2026.

[14] AICPA enrollment data reported in CPA Practice Advisor, August 2026: accounting enrollment at four-year institutions up 8.9 percent in spring 2026, the third consecutive year of growth.

[15] CFO.com analysis of the AICPA 2025 Trends report, October 2025, on AI absorption of entry-level accounting work relative to pipeline recovery.

[16] Zhou, X., and Ladley, D., “Integrating AI Strategically in Challenging Times,” AACSB Insights, March 2026, proposing the 4P framework of People, Policy, Pedagogy, and Platform.