The new drudgery

October 11, 2024 9:00 AM

We've all heard the claims by now: Artificial intelligence is going to completely revolutionize the accounting profession, handling all the drudgery, and freeing human accountants to do only the things that interest and engage them.

So is that it? Have we finally reached the apogee of accounting technology? Have we truly seen the end of boring, unfulfilling, unpleasant drudgery?

To find out, we posed this following question to a wide variety of AI thought leaders in the accounting solutions space–consultants, practitioners, vendors and academics:

Once we achieve our glorious AI future, where bots handle all the boring and unpleasant tasks accountants today are desperate to avoid, what boring and unpleasant tasks will come to replace them? When we finally solve the old drudgery, what will emerge as the new drudgery?

Their answers are below. Meanwhile, you can read some of my own thoughts on this subject here.

Danielle Supkis Cheek

Vice president, head of analytics & AI, Caseware

One relevant aspect that we could see in the future is the fact-checking process. As generative AI continues to provide better provenance of data in responses, including links, chain-of-thought reasoning, citations, and more, it will still be important to verify the accuracy of this information.

Randy Johnston

Executive vice president, K2 Enterprises

While AI will handle many transactional tasks, high-value and high-touch tasks will remain. Sometimes these can be exhilarating and pleasant, and sometimes they are unpleasant. Managing people inside the firm and interacting with clients will remain and will likely have both pleasant and unpleasant sides. Further, reporting bad news to clients, such as poor performance or fraud, will remain unpleasant.

Ellen Choi

Chief innovation officer and co-founder, Aiwyn

I think there is always going to be drudgery. First, there's those black swan events that AI cannot anticipate because there is no precedent for it, so you'll have to be kind of paranoid in watching out for things like that, the things that AI still can't handle.

This isn't that different from having an intern and having to micromanage what they do. An AI will always have to be managed. Hopefully it will get better over time, but you will always need to manage them, manage their output, to make sure they're doing what you want. And who knows, maybe over time if they become more like natural humans, it might be even more annoying to manage them from an emotional labor perspective.

Meanwhile, your AIs may need to also match the emotional tone of businesses you work with; maybe they want something more human, maybe less? Regardless, even when AI is doing all this work, managing and overseeing AI will never go away.

Abigail Zhang

Accounting professor, University of Texas at San Antonio

In the age of AI, we will have new types of drudgeries. The first potential drudgery is validating the relevance and faithfulness of the data we provide to AI. Although AI can automate the task of processing data, we need to do due diligence to avoid "garbage in, garbage out."

The second is data standardization. Fundamental AI systems like machine learning work with well-structured tabulated data. Since most raw accounting and auditing data are not naturally well-structured, pre-processing the data in a format that can be efficiently worked with by AI tools will become a new boring and labor-intensive task.

Another new drudgery could be prompt engineering. Since generative AI can be sensitive to the quality of prompts, the effort to test and come up with appropriate prompts could be a new type of labor-intensive task. Although there are applications that help the user craft a prompt, the user still needs to vet it or test it before simply relying on it, especially for accounting and auditing tasks. Of course, we will also spend more time validating AI's output and even auditing AI systems, which could become new drudgeries.

Wesley Hartman

Founder and CEO, Automata Practice Development

SONIA ALVARADO

With the progress that is being made in developing and using AI tools, it is tough to say what will emerge as the new task that everyone dislikes doing. In the short term, I think the new task will be reviewing what AI is generating. Accountants have been trained to verify data such as reconciling books, preparing tax returns or audit and attestation. Accountants would not want to put their stamp on work that uses their reputation without checking the work. Additionally, AI cannot be blamed for incorrect data because AI is not perfect, and a regulatory body will not accept "It was AI's fault."

In the long term, I think communications will become the new drudgery — which could be argued is a current drudgery. Email and chat have made us accustomed to instant transmission of our communications and the expectation of instant responses. When email came out, it was loved. No need to deal with slow faxes or couriers and electronic communications was efficient. Now, everyone has massive inboxes that are unmanageable, so we replaced email with chats, which started off great. Now chats notifications feel constant and the need to reply hangs over the head. AI is already being used to assist with communications such as drafting messages, but that goes back to my previous note on reviewing what AI is creating. Taking that to the next level, we will have AI systems instantly responding to AI systems, which can go off the rails with no value being generated for anyone.

Ben Wen

CEO and co-founder, Tallyfor

One issue will be who owns the liability for the work product. As a software vendor, I don't want it. But someone has to click the "Sign" button and accept it. What will likely happen is we will see the rise of click farms to hit that button for people, driven by regulatory requirements that don't keep pace with the technology. Think about how people get some drugs online: The necessary prescription for Minoxidil or Viagra is handled through a perfunctory Zoom call. Liability for AI might work in a similar fashion. Follow the insurance policies is another way to model it.

Also consider "artisanal" crafting. I used this line at the end of a donation pledge email: "*This email was organically, naturally, artisanally crafted without GPT, rBST, nor NFTs." We see in other markets that people are willing to pay more for handcrafted stuff over stuff made in a factory, even more if it's small batch and local. Maybe we'll see similar dynamics in the accounting world?

Another thing is, functionally, being a front-end interface for your grandma. Many people already know the experience of helping their aging parents navigate their MyChart medical emails. This is, effectively, being a front-end for the robots already. Even when AI is everywhere, not everyone will be able to use it without help at first, so users will also need to walk people through how to operate it like they do for their medical emails now. Some of it you could do over the phone/video call, but some will need physical visits to show people what to do.

And while I don't know the specifics, I'm sure that the resurgence of crypto will lead to some drudgery-like things to do, like how we see Minecraft and Roblox today generating tokens in a gamified process. Will people stand in some virtual line for an artificially scarce product? Will they need to undertake some skill-or-luck based process to increase their winnings or even chance of winning? I don't know, but it might lead to new drudge work no one wants to do, least of all accountants.

Blake Oliver

Founder and CEO, Earmark CPE

Blake Oliver

Before Excel, accountants manually created journal entries and crafted spreadsheets by hand, involving a lot of computation and requiring many bookkeepers. Since the invention of the electronic spreadsheet, the number of bookkeeping jobs in this country has declined by about 50%. However, we saw the creation of new jobs, such as financial planning and analysis. Financial analysts as a job category didn't exist before spreadsheets. If you combine that with financial managers, 1.5 million new jobs were created since Lotus 1-2-3.

So, what will happen once AI is creating the spreadsheets? We'll still need accountants, just as we still need bookkeepers, but we won't need as many. The drudgery-type work will likely center around data quality assurance — ensuring that what comes out of the AI is accurate and that the spreadsheets make sense. Those of us who have been managers know it can take just as long to review work as to do it in the first place. So, accountants will be reviewing the work produced by their AI agents. A typical accountant will supervise a team of autonomous AI agents performing discrete tasks such as accounts payable, generating journal entries, or bookkeeping.

In short, AI will handle data entry and much of the data manipulation, but human oversight will still be necessary to review the work. While it won't take as many accountants to do this, we will likely see new jobs created due to this technology, though we don't know exactly what those are yet.

David Wood

Accounting professor, Brigham Young University

As AI replaces more tasks, new tasks will emerge around maintaining these systems. This includes the often tedious work of data cleaning, model training, and fine-tuning. While critical for AI performance, these tasks may become the new "necessary evil." Professionals might find themselves spending hours sifting through datasets and adjusting parameters — work that, while crucial, could quickly become monotonous.

Also, we may soon see "AI auditors" whose role is to review and validate AI system outputs. This could involve meticulously checking AI-generated financial reports or tax filings for accuracy. While initially engaging, this role risks becoming repetitive, especially as AI systems improve. Constantly scrutinizing nearly flawless AI work, knowing that errors are rare but potentially catastrophic if missed, could lead to a sense of performing busy work. Eventually this type of work won't exist when the models are deemed "good enough," but until then, someone has to check the output.

AI's boost to productivity might also paradoxically increase workloads instead of reducing hours. As firms rush to leverage AI capabilities, we could see a surge in burnout and stress. The promise of simplified work may backfire, turning the profession into a relentless grind. Accountants might find themselves in a cycle of constant activity, blurring work-life boundaries. This intensification could transform even enjoyable tasks into drudgery, as professionals struggle to keep pace with expanded responsibilities. Tools meant to alleviate workload might create an environment where everything feels like an endless to-do list, leaving little room for job satisfaction.

In the longer term, we may consider perpetual upskilling fatigue: While learning is initially invigorating, the relentless pace of change in AI and automation might lead to burnout. This burnout could be caused by being trapped in a cycle of continuous learning, where newly acquired skills quickly become obsolete. This perpetual race to stay relevant could transform professional development from an opportunity for growth into a source of stress and resentment. The pressure to constantly adapt might leave professionals feeling like they're always playing catch-up, never able to fully master their domain before the next wave of changes arrives.

Carmel Wynkoop

Partner-in-charge of AI, Analytics & Automation, Armanino

Robert Houser

As AI takes over the day-to-day accounting tasks, ensuring data accuracy and the integrity of AI models/automated systems will become essential. The new drudgery will likely revolve around managing, monitoring, and ensuring the effectiveness and fairness of AI systems — an evolution of the profession that still demands human oversight and responsibility, just at a higher and more abstract level.

Accountants and finance professionals will have to manage and audit the data, algorithms, and their outputs to ensure compliance and correctness.

Even with sophisticated AI, there will still be outliers, edge cases, and unique situations that AI systems may struggle to handle. Humans will likely have to step in for manual oversight when the AI encounters exceptions or irregular transactions.

Someone will also need to maintain, upgrade, and audit the AI systems that perform the accounting tasks.

Accountants and auditors may be called upon to continually grapple with ethical and legal questions, such as how to handle AI bias, ensuring transparency in AI decisions, or safeguarding against systemic risks in AI-driven finance.

It's likely that accountants will spend more time communicating results to stakeholders, educating non-technical people on results, and managing client or board expectations around AI outputs being leveraged.

And while the nature of the work will change, it will involve a different kind of complexity and responsibility, requiring professionals to adapt their skills and embrace new opportunities in the evolving landscape of finance.

Brent McDaniel

Chief digital officer, Aprio

M.I.A Braids

AI will fundamentally transform the way we approach accounting. It will automate routine tasks, deliver improved predictive analytics, elevate audit accuracy and efficiency, and facilitate deeper, more insightful client interactions. This won't just make accounting firms more efficient — it will enable us to bring unmatched value to our clients. That's the promise, but it's a journey, not a single task to complete.

At Aprio, we're leading the charge by empowering our people and enhancing their skills to make them more efficient and, ultimately, even better accountants. Will AI change the daily role of accountants? Yes. Will it introduce new routine tasks along the way? Absolutely. But with the rapid pace of AI development, those tasks will soon be automated as well, moving us closer to realizing AI's full potential.