If AI takes on junior-level work, how will CFOs develop talent?
For the first 10 years of his career, Brian Beaupre built financial models in Excel and learned finance by developing the numbers that were behind major business decisions. It was how he developed his skills before eventually becoming CFO of Teikametrics, a company that makes software for e-commerce sellers.
“Quite literally, what these AI tools are doing today is what I did for the first 10 years of my career,” Beaupre said in a June interview with CFO.com at the CFO Leadership Council’s Spring Conference. At the time, his advice to younger finance professionals was to get comfortable with AI and understand the data feeding it.
A few months later, Beaupre is now thinking about what happens when the tools do too much of the learning for them. AI fluency, he said in comments shared on the topic with CFO.com, is on its way to becoming a standard qualification for new graduates. What concerns him is whether they will develop the judgment needed to know when an answer is wrong.
“That struggle is where judgment comes from, and judgment is exactly what you can’t automate,” Beaupre said.
The concern has changed how he evaluates candidates, mostly by spending less time testing technical proficiency during interviews, assuming applicants have similar skills in that area. Instead, he asks them to describe a moment of adversity or a problem they faced without a clear answer.
“How someone tells that story, and how self-aware and honest they are about what they got wrong along the way, tells me far more about their ceiling than any technical screen would,” he said.
The experience behind the answer
Beaupre’s approach raises a question that extends past hiring: If AI can complete the work that taught previous generations of finance professionals, where will their successors gain the experience to check it?
Digits, an AI-powered accounting platform, had its co-founder Jeff Seibert address part of that question during an appearance on The Accounting Podcast that was recorded the day after Beaupre’s interview with CFO.com in Boston. When asked what a new accountant should learn to remain valuable as the work changes, Seibert pointed to the decisions that arise after a tool produces an answer.
“You need someone with lived experience who is able to guide it and make those difficult judgment calls,” Seibert said. Accounting work, he added, includes decisions that cannot be resolved by following a clear rule every time.
“Folks that understand the space, the complexities, and have the experience to make those calls, that is what you’ll spend a lot of time doing,” he said.
Seibert’s description fits the skills Beaupre wants to find in an interview; a candidate who can explain how they handled uncertainty may offer more insight into their potential than one who can complete a technical exercise. Whether Teikametrics has already seen AI interfere with the development of junior employees is less clear; Beaupre described it as a risk that concerns him.
“What actually keeps me up at night is something upstream of that: the risk that leaning on AI too early in a career erodes the muscle of critical thinking that only gets built by struggling with a problem, getting it wrong, re-learning it, and eventually solving it yourself.”
For Seibert, the human role also comes down to responsibility for the result. “The AI can never take accountability,” he said on the podcast. An accountant still has to build a relationship with the business owner and stand behind the numbers they provide.
Those are the exact elements of AI that are putting pressure on finance leaders like Beaupre, mostly around the way junior employees are being trained. Reviewing an AI-generated result requires two key traits that are often developed in the gritty work of finance and accounting — enough prior knowledge to spot a questionable assumption and enough confidence to challenge it.
An experienced team becomes harder to replace
The value of that experience is already apparent in hiring discussions. On the episode of The Accounting Podcast that Seibert appeared on, host Blake Oliver later described a senior accountant as the most difficult role to fill, a finding from a recent CFO survey by Personiv he discussed on the show, followed by staff accountant. He connected the shortage of senior talent to the difficulty firms have had keeping people in the profession.
“Staff accountants are deciding, after a couple years, ‘I don’t really want to keep going in this profession,’ so they’re dropping out,” Oliver said. “That’s why it’s hard to find seniors, right?”
AI may make that experienced layer more valuable if firms need senior employees to direct automated work and review the results. As Oliver put it, “Recruiting is now as important as business development” for an accounting firm whose work still depends on people, even as it uses AI.
Oliver also discussed evidence that companies continue to hire at the start of the career ladder. “Accounting hiring for recent grads is flat,” he said, while hiring for many other white-collar jobs has declined. He cited Bennett Thrasher as an accounting firm that has kept its entry-level hiring in place.
At Teikametrics, Beaupre said he is investing in AI and automation training for tenured employees. He added that the approach varies because established staff have different ways of learning and different reasons for trusting, or questioning, a tool’s output. His team is finding that some need repeated practice, and others need to see AI applied to a problem they already understand.
However, their existing habits carry knowledge Beaupre wants to retain. Pairing those employees with newer hires who are comfortable using AI could help each group learn from the other, he said. Experienced staff can show how they reach a decision, and new employees can help them work with tools that are becoming part of the job.
The arrangement depends on balancing learning with getting work done and all the risks that are associated with that. But if an employee sees only the finished AI-generated model, they may miss the assumptions that made it useful or led it astray. A senior colleague who explains those choices can give the new hire a way to assess the next answer for themselves.
“The AI and [robotic process automation] tools are the easy part,” Beaupre said. “The differentiator is building people who can recognize a problem worth solving, stay humble about what they don’t know, and earn the trust of the people around them as a real business partner.”
