Mind of Murph: What's Next for Accounting

One thing is becoming obvious... Accounting as we know it is changing.

Will it change faster than we anticipate?  I now believe the answer to that question is, "Yes."

Will it change for 'the better'?  I now believe the answer to that question is, "Only Somewhat."

To explore these questions in greater depth, let's look at a 'brief history of accounting'. 

Early accounting records have been identified in ancient Mesopotamia where clay 'tokens' were used to keep track of crops, herds and traded goods. We also know the ancient Egyptians had a rudimentary form of accounting to keep track of goods, primarily grain, moving in and out of storehouses. Rome, in the time of Caesar and his successors, had well maintained records of expenditures approved by the Emperor or Senate.

Medieval Italy seems to be the birth of at least one form of bookkeeping; although it was in the late 1400's that an Italian mathematician, Luca Pacioli, wrote the first 'accounting textbook' (of sorts), detailing the use of debits and credits for each transaction. His works led to the general acceptance of double-entry accounting in Europe by the mid-1500's. 

Yet most 'bookkeepers' still recorded business transactions in a single-entry fashion with a quill pen on blank paper, as did Bob Cratchit, by the light of a single candle, for Ebenezer Scrooge in Charles Dickens' Christmas Carol. It really was the late 1880's when the accounting profession was born. Bookkeepers for a 100 years made double-entry journal entries on columnar pads that were put into post-binded books.

Then Apple launched VisiCalc in 1979, followed by Microsoft's SuperCalc in 1980, and Lotus 1-2-3 in 1983. Of course, if you happened to have access to a main-frame computer, you might have used Lanpar 'spreadsheet' accounting as early as 1969. Late in 1985, Microsoft launched Excel which is still considered the 'spreadsheet of choice' for those using spreadsheets. 

My point is simple, someone, or some thing, changes the accounting industry every so often.

Today, everyone is talking about Artificial Intelligence changing Accounting. It's already happening, but the question everyone really wants to know is "will AI change Accounting for the better?" 

Time was, tens, if not hundreds, of individuals worked in large businesses across the globe 'doing the number crunching' for everything from tracking the accounts receivable to the flight trajectories for the first manned space flights. Human computers did the math that most individuals were never taught, and it took huge offices full of these individuals to perform computations that today we can do on our cell phones. 

Then, in the late 1950, early 1960, tools like the Friden Model STW-10 calculator came along and it could literally 'crunch the numbers' in minutes as the 'stepped reckoner' wheels spun. Even nine cycles of the machine to compute a single solution took only a short time. The result... what had taken nine individuals to accomplish, the machine could now do. Eight fewer jobs, but it was still called progress because it took far less time to accomplish the end goal. 

The story Hidden Figures₁ tells of the human computers who performed crucial calculations by hand, and by use of Friden (or similar) mechanical calculators, to compute the essential numbers for both sub-orbital and orbital flights of Project Mercury, the first US manned space program. By the end of Project Mercury, an IBM 7090 Processor was providing the raw computational power for the critical calculations needed for the Gemini spaceflights. In so doing, computing technology eliminated a vast majority of 'human computers' working for NASA and that was considered 'progress.' 

In the same way, the advent of those computer spreadsheets eliminated the need for multitudes of bookkeepers who were responsible for specific 'journals' within a business, each of which were cumulatively recorded by additional bookkeepers into a single general ledger. With applications like Excel able to 'maintain the entire accounting architecture', the number of bookkeepers was reduced substantially for big business.

Growing up, I saw this happen in my dad's small engineering firm. Prior to buying a Friden calculator my dad had two engineering techs who computed one figure after another in taking off the costs for a new project. One day, I asked my dad where Larry and Daniel were and he told me he 'laid them off' because the new calculator could do in 15 minutes what the two men had taken 2 hours each to do. As best I recall, that was my first "life experience" with technology replacing humans. 

Clearly, just as the computer age replaced humans, artificial Intelligence will replace accounting personnel. Yet, that technology can accomplish a wealth of tasks and functions that accountants have had little time to perform. The proliferation of 'Agentic AI' is streamlining processes and performing analytics in ways that few small accounting systems have been able to accomplish, at least not without the aid of other Apps. 

Now you are asking the 'big question', will AI replace "You"?  That depends on who "You" are. 

If you are at or near the bottom, or even the middle, of the 'Accounting food chain', the answer is likely "Yes, AI will be replacing you." And, the reality is that you will probably be replaced in the next couple of years, if not before.

For example, if you are an Accounts Payable clerk for an accounting firm, or a small business, your days are numbered. AI-agents can directly import a vendor's invoice into the accounting technology, code it based on instantaneous analysis of coding patterns for the same vendor, item and payment history. The AI-agent can then  electronically solicit approval from every level of authority required prior to sending electronic payment consistent with the vendor's choice. So where is the need for all those A/P clerks? In simple, there isn't a need. They soon will be, if they are not already, on their way to an early retirement. 

If you are an accounting tech who currently 'prepares' financial statements, "watch out"- your job is likely to be the next that AI wipes from the planet. Accounting statement preparation is designated as 'Prepared,' 'Compiled,' 'Reviewed' or 'Audited.' Each has a progressive level of complexity and 'assurance' associated with them. AI-agents are already 'preparing' period-end financial statements meeting the requirements of statement assembly in proper financial reporting format. AI is even preparing commentary that mirrors 'notes to the financial statements.' 

At the next level of accounting statement preparation, 'AI' may not yet have the edge over humans. Even though 'compilations' essentially express no level of assurance or opinion regarding the financial statements as a whole; most businesses take at least some level of comfort in knowing their 'accountant' has made his (or her) self familiar enough with the statements that they are not aware of material mistakes.

Only time will tell if businesses will be comfortable with 'AI' performing true financial statement compilations including a transmittal letter. Furthermore, Accountants who wish to rely on 'AI' prepared compilations may in fact find that they spend more time 'reviewing' such compilations so as to resolve misstatements prior to submission to the client. 

And who is the client going to trust?

For at least the last ten years the term 'trusted advisor' has been batted about like a tennis ball. The emphasis was on turning the relationship of bookkeepers and accountants into more of a 'counselor' than a 'number cruncher' who sends you a report.  As part of that emphasis a renewed interest in the computation of key performance indicators (KPIs) and business metrics for each client became a focal point of 'starting the conversation' between professional and client. 

While I am first to admit that computing KPIs is not only difficult but time consuming, and the same can be said for 'metrics' based on the financial records, they can be invaluable to businesses for purposes of cash flow forecasting, business performance trending, and financing opportunities. The fact that AI-agents can compute, trend, and develop presentation graphics and models of this information in the blink of an eye propels the ability of the accountant to provide this information to the client, even with a grammatical analysis of what the data means.  

But as most of us know, most people in business want a two to four hundred word 'Management Overview', not 30 pages of content to read. And if they have questions they want 'to the point' answers, not lengthy answers including question restatements.

So, do we really think that a client is going to 'listen to' an Agentic Accounting Agent 'Siri-speak', "your debt two asset ratio is two hi at two two two one₂."  AI (or at least Siri) is not quite ready for 'prime time.' 

It probably would only take one episode of a client being exposed to that nonsense to quickly determine the firm's new role as, "not my trusted advisor any longer."  That means, for the near future, you had best 'shield your clients' from exposure to AI interpretations of their finances.

You may have help through regulation.

Let's not forget that 'accounting' is a profession, and what makes it such is that it is  self-regulating (for the most part). While states codify rules that license accountants, and the IRS mandates rules that define regulations, these and the other various regulatory bodies governing 'accountancy' are composed of accountants who seek to regulate the profession for the sake of the profession (as well as the 'protection of the public.') 

At what point these regulatory bodies say, "Enough is enough of this AI stuff. You can't use AI for performance of 'compilations," or "Under no circumstances will AI be used to perform any 'financial statement reviews' or 'financial audits.' It could be that this type of regulation of AI within the accounting industry is not far off at all. 

But the soonest response may come from a different form of regulation. Professional Liability Insurance carriers who write coverage covering Accountants, with specific policy riders for 'reviews' and 'audits.' They may quickly come to realize the added liability exposure they fact if AI is preparing, compiling, reviewing and in fact auditing every phase of the accounting process. They will likely either announce they are no longer going to write policies where the insured is using AI (or is using AI for specific accounting tasks), or they will raise their premiums to the extent that accounting firms can not afford to purchase coverage if they use AI, and with such controls in place, the use of AI will likely come under much stricter limitations. 

But who know, the AI-bots may very well take over the insurance business and become the actuaries and underwriters of the next generation and decide to only write insurance to AI-controlled accounting firms.

So, how long will it be before Agentic Agents put human 'face forward' representatives of the accounting profession in their rocking chairs?

The answer, "only tech-time will tell."  Some people will have you believe that we can 'master the technology' rather than it 'mastering us.' While that may be true in the short run, I doubt it is a viable 'long-term play.' Because 'tech-time' moves really fast. The AI-Bots are absorbing knowledge faster than the speed of light. They are actually learning from the AI tools that editors are using when reviewing manuscripts. That means the bots know the book or manuscript content before humans even have the opportunity to read the published book. 

Some argue, but the Bots don't know how to apply that knowledge until they are taught by humans (aka: 'trained' to use it). Bots are already training other bots, coaching other AI-Agents to do the tasks based on millions of computations as to what is possible and what is most probable. And when in doubt, well they propagate their own rules until someone comes along and changes them, like when the AI-Agent codes your transaction this way, but you decide that's wrong. 

But how long before the AI-bots and agents decide, "who is this human to tell us that we are wrong?" 

I suspect, when the AI-Accounting Agent of 'XYZ accounting firm' is communicating directly to the AI-CFO Agent of 'Big Business ZYX', the role of humans in the accounting loop will become faded pictures in the human archive and stories within obsolete history books.

What will 'accounting relics' like us be worth when AI-agents have become the 'great white sharks' of the accounting food chain? 

In 1960, a new Friden mechanical calculator cost $880, just about the same amount as the monthly salary of a 'top level' accountant. Today, if you can find a Friden calculator in working condition, you can likely buy one for $250 (28% of the original cost).

What do you think? Let us know by posting a comment at the bottom of this article. 

Footnotes: 

₁ - Hidden Figures is a 2016 non-fiction novel by Margot Lee Shetterly about three female African-American mathematicians who worked at NASA during the early space race. The novel was loosely developed into a film by the same name, produced by Fox 2000 Pictures, Chernin Entertainment, and Levantine Films, and release within the U.S. in 2017 by 20th Century Fox. 

2 -"Actual Siri quoted SMS text message from an AI-generated query response."