Brace yourself for AI and blockchain
At first glance, the threats seem clear: One type of software will learn how to perform all manner of business functions, particularly in finance and accounting, while another will continuously validate any set of data or information.
Between them, artificial intelligence and blockchain seem poised to disrupt — or even destroy — many of the core businesses of the accounting profession, automating or rendering irrelevant important traditional services like the audit. But while there can be little doubt that they will eliminate the need for human beings to perform many of the individual functions traditionally associated with accountants, both in public practice and in industry, they will certainly not eliminate the profession’s overall role, or its importance.
In fact, both AI and blockchain have the potential to help accountants actually boost their revenue, their relevance and their value — provided they’re willing to develop the necessary skills, and change their mindsets.
Understanding why each of these two emerging technologies is less of a threat and more of an opportunity than they might seem requires a separate, deeper dive into each, as they’re going to have different impacts on the profession, over different time horizons.
It’s only partially accurate to describe AI as an emerging technology; it has already emerged in some forms and some applications. It’s at the core of IBM’s Watson, for instance, which Big Four firm KPMG is applying to its professional services offerings, with a focus on auditing. H&R Block is adding Watson’s artificial intelligence to its tax prep process, while the Maryland Association of CPAs is working with IBM to train accountants in technology skills like AI and cognitive computing.
All this begs the question: What, exactly, is AI? Pop culture gives us HAL from “2001” as one example, Skynet from the “Terminator” movies as another, and Scarlett Johansson’s disembodied voice in “Her” as another, friendlier one, but those are fictional characters, not models for AI. “A computer that thinks like a human being” comes close to the overall goal of the field, but that’s an idea that’s both nebulous and fairly far off (to say nothing of not necessarily worth pursuing, given the quality of most human thinking).
A more useful definition, and certainly one more in keeping with the current state of the field, is that AI is software that can draw conclusions from large quantities of data, and adjust its activities based on those conclusions — that can, in effect, learn. Leon Katsnelson, director and chief technology officer for strategic partnerships for data science at IBM, cites the example of an elevator company having AI go through reams of data from all of the sensors on its individual cars, identifying from that data the characteristics of an elevator car that is about to have problems. The AI can then keep an eye on all the company’s elevators going forward, and dispatch maintenance crews as soon as they exhibit any of the pre-problem characteristics.
Essentially, AI is about software that can learn and adapt, which is why one subset of it is called machine learning. “Machine learning is like a rocket engine and data is the rocket fuel,” Katsnelson explained. “In traditional programming, we’re teaching the machine how we do the job — we’re telling it, ‘Repeat what I do.’ AI is about teaching the machine to learn how we learn — to learn from data.”
That means that it can learn without human input, and that it can act without human direction. It can also analyze far, far greater amounts of data than a human being ever could — and make useful decisions and recommendations based on that data. Given all that, it’s not hard to see why it’s considered the next big thing.
