As AI-generated text becomes easier to identify, accounting and finance graduates may need more than polished resumes and academic credentials to prove what they can actually do. Manuj Mittal, A former finance and accounting professional turned higher education researcher, has spent years developing an approach centered on applied skills and workplace readiness.
Sometime soon, a recruiter will open two nearly identical cover letters. One will carry an invisible watermark showing the text came from an AI model. The other won’t.
The harder decision begins after that.
The future of work moved closer last week, when Anthropic explained how it will embed invisible marks in text generated by its Claude models. The company is using a version of SynthID-Text, an open-source system developed by Google DeepMind that works through word choice. When several words fit a sentence equally well, a hidden key settles which one appears. Repeated across a passage, those choices form a pattern that detection software can read and no human eye can see. Anthropic is complying with the European Union’s AI Act, which requires machine-readable marks on synthetic content, and Google’s Gemini has carried text watermarks since 2024. Other AI providers operating under the EU’s rules face similar transparency requirements.
The technology might sound like a technical footnote until you consider who relies heavily on AI-generated writing. Students are already among its most frequent users. In the UK, 92 percent of full-time undergraduates now use AI tools and 88 percent have used generative AI for assessments, according to the Higher Education Policy Institute’s 2025 survey, while a working-paper survey at Middlebury College found 82.5 percent of American students use generative AI for schoolwork. The habit doesn’t end at graduation. Seventy percent of job seekers use generative AI to research companies, draft cover letters, and prepare talking points, according to Indeed’s global AI survey.
Watermarking won’t catch all of it. Detection requires a key, and text can be edited or run through tools that carry no marks. But the assumption that machine writing blends invisibly into human work now has an expiration date.
Watermarks answer one narrow question: the origin of the work. They don’t tell an employer whether the person submitting it understands the underlying material.
Manuj Mittal has watched that gap from both sides. After a career in audit, advisory, and investment finance, including roles at Grant Thornton and Morgan Stanley, Mittal shifted into graduate education at the University of Rochester’s Simon Business School, where he has prepared finance and accounting students for a profession rapidly changing under AI.
His career has put him on both ends of the transition. He has evaluated young professionals as an employer and prepared them as an advisor. What he sees now is a profession whose traditional entry point is changing.
For decades, junior accountants learned the field through work few people outside the industry noticed: reconciling accounts, testing journal entries, sampling audits, preparing first-pass tax returns. None of it was glamorous, but it taught new hires how the numbers fit together. Much of that routine is increasingly automated. What remains is what machines still struggle to do: professional skepticism, client relationships, recognizing which anomalies actually matter.
“Smart graduates can produce polished documents, but the differentiator will be whether they can demonstrate sound judgment, defend their reasoning, and apply what they know when there isn’t a clear answer,” Mittal said.
The industry’s predicament, as Mittal describes it, is a strange math problem. The U.S. workforce lost roughly 340,000 accountants and auditors between 2019 and 2023, according to Bureau of Labor Statistics data cited by Bloomberg, and first-time CPA exam candidates fell 33 percent between 2016 and 2021, according to the AICPA’s Trends Report. The profession needs more people at exactly the moment it needs different people.
“The professional world needs more accountants, but it also needs accountants prepared for a different kind of work,” Mittal said.
His answer is to change what happens before graduation. Mittal argues that experiential learning belongs inside the curriculum itself, not bolted on as optional internships or career-center workshops. That could include employer projects, integrated work placements and simulations that mirror real workplace decisions, such as managing a quarterly close for an employer partner.
Assignments built around real workplace problems give students a chance to demonstrate how they apply what they have learned before they enter the profession.
“Students shouldn’t graduate with only grades and résumés. They should leave with evidence of what they can actually do, how they approach problems, and how they perform in situations that resemble the workplace,” Mittal said. “That experience gives employers a much clearer picture of whether a graduate is ready for the work.”
Coursework built around real employer problems produces a record of capability that exists outside any document. A student who ran a simulated close, presented the findings, and defended them under questioning has evidence no watermark can grant or revoke. No detection key is needed to verify it.
Mittal has already built one version of this at Simon Business School, where he helped design and co-instruct Career Strategy and Professional Readiness, a 14-week graduate (credit-bearing) course that puts students in front of employers before they ever enter the hiring process. Judgment and communication, in his experience, develop fastest when students practice them in real settings rather than reading about them.
He is not arguing that universities should abandon AI, or wall students off from it. Employers increasingly expect graduates to work alongside these tools rather than simply perform the routine tasks the tools now complete, and Mittal believes preparation for that reality belongs in the curriculum as much as debits and credits do. “Keeping AI out of the classroom prepares students for a workplace that no longer exists. The job is to teach them to work alongside it while building the skills it can’t replace,” Mittal said.
His doctoral research at Rochester’s Warner School of Education pushes the idea furthest, examining how universities can redesign graduate programs themselves to align academic preparation with a fast-changing profession. The focus is institutional: changing programs, rather than advising students through them one at a time. Some educators argue universities should prioritize broad intellectual development over workforce preparation. Mittal believes both are essential, and that upskilling younger generations strengthens the communities they go on to serve. “In the corporate world, AI and other emerging tools are already changing how jobs are designed. As community builders, we need to start by reimagining education,” Mittal said.
There are early signs his side of the argument is gaining ground. Accounting enrollment grew 12 percent year over year for two consecutive semesters in 2024-25, according to National Student Clearinghouse data cited by the AICPA, and model licensure changes approved in 2025 created an additional pathway to CPA licensure that states can adopt without the traditional 150-credit-hour requirement.
Increasing the number of candidates, however, does not address the larger question raised by AI watermarking. Invisible signatures may eventually reveal where a document came from. They still cannot explain how someone thinks, how they make decisions, or how they perform when the answers aren’t obvious. Those are the qualities employers have always valued most, and the ones Mittal believes universities must do a better job of making visible before graduation.
“Technology will continue to change how accountants work. What won’t change is the need for professionals who can exercise judgment, earn trust, and solve problems that don’t come with an answer key. Employers will always be looking for candidates who can demonstrate those abilities,” Mittal said.
