Who Teaches the Accountant When AI Does the Work?

Editor’s Note: This article is an opinion piece. The views and opinions expressed are those of the author and do not necessarily reflect the views, positions, or policies of Woodard.

I was pretty excited when I finally got AI to take over a big chunk of the daily sales work for one of my restaurant clients. I used to take the POS report, enter the information into Excel, make sure everything footed, create the import, and check that the cash and credit cards made it to the bank. Now AI reads the report, creates the import, puts the numbers in the right categories, and does most of that work for me. Magical!

This is exactly what I want AI to do. Then one day it processed $14,810.34 in retail sales. This client averages $270.57 a day in retail sales, so $14,810.34 is a lot of T-shirts. AI did exactly what I asked it to do. It took the number from the report and put it where it belonged.

What AI did not do was stop and say, “Hey, this looks wrong.” I did.

The work was teaching us

Those of us with experience often know something is wrong before we can explain why. We just feel it. It is like that slightly crooked picture hanging in the doctor’s office. It is off. You see it and wonder why no one else has.

Where does that feeling come from? It comes from doing the work. Back in the day, I sat at a desk with green ledger paper and a sharp pencil and an adding machine on my desk spitting out long trails of paper that were stapled to the workpapers as proof. There was no undo button and certainly no AI working overnight while I slept. If something did not balance, I had to find it.

I got things wrong too. I had to go back and figure out why – my own version of Indiana Jones. I am certainly not asking to bring back green ledger paper. (Although I have to admit, there are pads of them in my office. I cannot quite bring myself to throw them away.) All that repetition taught us what made sense. It was how we learned to recognize what was off, when things made sense, and the entire accounting cycle.

The work that took us hours is disappearing. Much of it should. But our learning was buried inside that work.

Reviewing AI is not a training plan

I keep hearing that AI will do the work and accountants will review it. That sounds fine until you ask what a new accountant is reviewing if they have never seen enough good work, bad work, and strange work to know what deserves another look.

My POS system knew the number. AI knew where to put the number. I knew the business. Those are three different things.

The software had no reason to question $14,810.34. That was the number on the report. I questioned it because I had seen this client’s numbers over and over again. I knew $270 was ordinary and $14,000 deserved my attention.

Experience gives you something to compare the answer to. Without that comparison, reviewing can easily become confirming that the software did what it was told to do. Reviewing is not the same as exercising judgment.

We have to let people think

Maybe part of the answer is simpler than we are making it. Stop giving people the answer before they have had a chance to look.

Give a newer accountant the financial statements and ask what they see. Do not immediately explain the unusual balance or point them toward the problem. Give them time with it. They may notice something important. They may spend ten minutes chasing something that turns out to be perfectly fine. Both teach them something.

Let them dig into it. AI can be terrific here. It can compare periods, trace transactions, and help find where a number came from much faster than I ever could with my green ledger paper. The difference is that the accountant started with the question.

I do this myself. I look before I ask AI what it thinks. AI catches things I miss too, and when neither of us has enough information, I ask the client. For me, that is where judgment gets exercised.

New accountants also need to hear client conversations. Bring them into meetings so they can hear what gets asked when something does not make sense. Eventually, they need to ask some of those questions themselves.

They are going to get some of it wrong. Good. I got things wrong too. Having to go back and figure out why was part of how I learned.

Make the time worth something

AI is giving me hours I used to spend processing data. I have no desire to spend those hours re-creating the work the technology just eliminated.

I want to use them to understand the business better.

Why did retail sales jump? Why are beer sales dropping? What were our sales in prior years before Bike Week so we know what to order this year?

That is what I want to spend my time thinking about. And that is what I want the accountants coming behind us to learn how to think about.