Can accountants build an AI workforce?
In short: Yes, they can
August 2019 Footnote
Editor's note: Updated July 31, 2019
Rumors abound that artificial intelligence (AI) will replace accountants and auditors — in short, there will be a robot revolution.
These rumors are unfounded as true
CPAs provide services that reach well beyond what can be automated. However, accountants can leverage an AI workforce to reduce their workload and enable them to focus on higher-value services.
Determining what AI can and can’t do
To understand what AI can do, use the Automation Framework.i
This framework is a 2x2 matrix with Complexity on the y-axis and Repetitiveness on the x-axis (see diagram A below). Using these two dimensions, you can map out the tasks to be performed. For example, the tasks performed by a retail salesperson are mapped out in Diagram B, or the tasks performed by a medical doctor are mapped out in Diagram C.
Diagram D then shows which types of automation technology can be used to automate these tasks. High-level programming (meaning you don’t need AI for this) can be used for a lot of things that fall into the Routine & Simple quadrant. Then the AI technologies of machine learning and deep learning can be used for the low-complexity and high-repetitiveness quadrants. This leaves primarily things that have a low repetitiveness and high complexity to humans.
Which accounting roles can be automated?
If you’re curious about how that plays out when mapped to the various roles within the accounting profession, you can check out the AICPA’s estimates based on Bureau of Labor Statistics (U.S.), McKinsey and Oxford Data, shown in Diagram E.ii
Essentially, most of the bookkeeping and clerk-type jobs can be automated using current technologies. This can already be seen with both QuickBooks Online and Xero, where both incorporate AI into their transaction coding processes, and vendors like Botkeeper whose business model is based on customized AI-enabled bookkeeping. Audit tool vendors like ACL, IDEA, Info and MindBridge AI all have incorporated AI into their applications in supporting audits. CPA.com also has a software called OnPoint PCR, which uses AI to help guide an auditor in creating the workplan for preps, compilations and reviews (PCR). Both consumer tax programs (e.g., TurboTax) and practitioner tax programs continue to improve their ability to lead a user through the complexities of the tax code and increase the accuracy of the tax return preparation process. It’s quite evident that AI is already helping to automate quite a bit of accountant and auditor tasks.
AI tools are accessible to and affordable for accountants
If you can’t find an application that directly automates a task, the technologies to create your own AI workforce are also affordable for accountants even at small firms or businesses. Microsoft, Amazon and IBM all make their AI technologies available via their cloud platforms, and there are a variety of smaller vendors that provide solutions that require varying levels of technology expertise.
A great part about the maturation of AI tools is that you don’t need a high level of technical expertise to use these tools anymore. This was perhaps best exemplified by the Chatbot Invention Challenge hosted by the TXCPA Houston and TrueUp, where accounting students built a variety of chatbots to handle routine interactions supporting order entry, accounting inquiries and other tasks using a tool where programming wasn’t required to automate conversation-based interactions with a human.
Tech-savvy accountants could probably easily leverage chatbots and other tools to automate some of their more routine and less complex work. However, it’s probably not the best use of the average accountant’s time to build these AI solutions. For most accountants, it’s probably wiser to find a consultant who has experience in building these solutions and to describe to them what needs to be done.
Accountant skills that help build AI workforces
Luckily, in our accountant skillset, we have a few competencies that lend very nicely to supporting the building of an AI workforce. Two primary competencies that are needed in designing AI solutions are business analysis and data analytics.
Business analysis is the documentation of a process’s workflow and processing logic. It also includes modeling the underlying data structures and relationships within the data. Both are critical to describing the tasks that must be automated as well as the underlying data that is consumed, processed or outputted as part of the task. Use these competencies to describe to a consultant what and how something should be automated using AI or even just conventional programming.
Data analytics is the analysis or interpretation of data, its trends and the drawing of insights or identification of anomalies or exceptions in the data. Strong knowledge of statistics and advanced analytical techniques also often come into play in this area. Many times an accountant can manually perform the analytics in Excel, which can then be used to show a consultant what needs to be automated, including acquiring the data from the source system and describing how to deliver the results to the person who will be interpreting the information.
Accountants must address data biasing risks
When implementing AI-enabled machine learning, accountants also need to ensure that data biasing risks are addressed. This is because AI learns from the data that it is provided. Data biasing can have both negative and positive effects.
For example, let’s say you wanted to use AI to predict which candidates you’re considering hiring would most likely become CPA firm partners; if you fed the AI all the demographic information about CPA firm partners, it would ultimately predict that white males would be the best candidates. That’s because today’s CPA firm partners are predominantly older white males — the data is biased toward that demographic. To prevent both racial and gender biasing (both Equal Employment Opportunity Commission issues), you would need to ensure that the AI was prevented from using those two data points as part of its predictive analysis.
Data biasing can also be positive. For example, Amazon uses this to predict what complementary items you may purchase when you put an item in your cart. It then suggests these items to you, hoping to increase your purchase while also anticipating your potential needs.
Sometimes, you may also need to identify situations where AI should not be applied because there is too much variability. For example, the deductibility of food and beverage expenses may vary depending on the business purpose of the expense. Unless the additional data elements are present that would dictate the proper categorization of the expense, accountants may want to ensure that an AI-powered transaction classification solution marks all food and beverage expenses for review by an accountant.
Accountants should be leveraging AI
Accountants can and should leverage AI-based tools, and even build their own custom AI workforce.
Look at your workload and use the Automation Framework to figure out what parts of your workload could be shifted to AI. Look at the tools already available in the market and see if they can be incorporated into your practice. Simple and common client interactions and inquiries, while easy for you to address, consume valuable time that, if handled by a chatbot, may allow you to focus on higher-value tasks.
If you think of AI as something that can help you, rather than something that you’re competing with, you’re sure to come out ahead and not be replaced by the robot revolution.
Donny C. Shimamoto, CPA.CITP, CGMA is the founder and managing director of IntrapriseTechKnowlogies LLC, a Hawaii-based CPA firm focused on innovation acceleration and risk management for small businesses, midsized organizations and nonprofits. He has been recognized many times as a Top 25 Thought Leader and Top 100 Influencer in the accounting profession.
i Source: Abhas Gupta, MD, “AI’s Threat to Society is Scarier Than Trump,”
https://medium.com/@abhasvc/ais-threat-to-society-is-scarier-than-trump-ff7e9d42ea74, Aug. 9, 2016
ii Source: Alfonso Olaiz, Lead Manager - Strategy, AICPA, presentation to National Conference of Lawyers
and CPAs, December 2017