One of the key advances in computer technology in the last ten years has been the transition from 32-bit to 64-bit addressing. While the technical ramifications of this shift are a fascinating topic for programmers, the practical benefits had largely escaped the average computer user until the advent of data mining for professions like accounting.

The best metaphor is to compare an accountant who can read and understand a single page of your company’s ledger at a time vs. the accountant who can read and comprehend all the pages at once. That’s the difference between a 32-bit and a 64-bit application.

When combined with sufficiently advanced pattern matching, electronic accounting systems began drifting into the realm of artificial intelligence. The question is, how do these systems compare with a qualified and experienced accountant?

Rows and Columns

From a technical standpoint, there is no difference between a spreadsheet and a database aside from the spreadsheet’s innate mathematical functions. All of the information is arranged in rows and columns. A pattern matching algorithm brings something new to the process, however.

One might think of it as a chess game. The chess computer’s objective is to checkmate the king. The accounting computer’s objective is to make the number on the last row, or bottom line, as large as possible within a certain set of rules.

The reason the accounting computer is so good at its job is that it can do math very fast — to the tune of billions of calculations a second. With the right inputs, an off-the-shelf laptop could do every tax return in the country in a matter of hours.

Multiple Tiers

When this kind of calculation engine is combined with any set of data, the task of optimizing that data becomes a standard function. Since accounting is governed by a common set of rules, making that optimization objective match the rules becomes a simple matter of arithmetic and pattern analysis. However, that doesn’t necessarily make the artificial intelligence better. It just makes it predictable.

The Human Factor

What humans do that computers cannot is intuitive pattern matching. Human beings can discard fantastically large sections of a probability matrix at a glance. A child of five knows lions aren’t blue. The five-year-old can, therefore, discard every premise in the probability matrix that leads to “lion” if the animal they are viewing is blue. For a computer to perform the same function, it would have to work through the entire set of possibilities and discard those that produce bad results one at a time.

However, when a trained accountant and a computer join forces, they can do both at the same time. The human can set aside all but the most important data, and the computer can train all its power on only the data that will produce the desired outcome.

Can computers replace accountants? In certain very limited circumstances, they might. While artificial intelligence can answer questions like “what?” and “how?” it takes human beings to answer the overarching question of “why?”