ChatGPT and generative AI are not the answer in and of themselves but are tools to help humans find the answers faster.
It's not surprising that as AI models get smarter, they can start passing some standardized knowledge-based tests. While passing a test may say you are a CPA on paper, you do not have any of the experiences and expertise developed over a career. Few would argue that someone who passed the CPA exam is immediately an expert on par with a CPA with 20 years of experience.
Without human oversight, generative AI can quickly run amok with inaccurate analyses and perhaps dangerous answers, especially when related to the tax code or an accounting method.
While the CPA exam does include what test-takers perceive as task-based simulations based on real work problems, they are not as focused on the unending scenarios the tax code actually leads to in the real world.
Some have commented that ChatGPT's performance on a practice CPA exam in a recent study calls into question the competitive advantage of the human accountant relative to the machine. This response shows a total disconnect from what accountants and tax professionals truly do for their clients. In reality, this job is much more about planning and advice, and even building relationships than it is in answering, "What is an income statement?"
When it comes to generative AI, one of the primary issues at hand is who is watching the machines? What controls are in place? Is the data accurate? What KPIs matter most? This is learned from experience, not textbooks. The researchers themselves had to train the machine on how to learn. It didn't accomplish this on its own.
The future is a combination of human fact-checking and generative AI responses. This is how the industry can ensure accurate data and information, without sacrificing the convenience that generative AI can provide.
One skill all tax and accounting professionals can master is writing prompts. Prompts help you feed generative AI algorithms the data and criteria you want to analyze. Once this information has been loaded in, the machine can summarize a topic, write a client email response, predict the future based on inputted data, and suggest improving results. However, the "garbage in, garbage out" concept applies here — if you are feeding the algorithm bad information, poor answers and content will be the result.
Consider this: How do you decide what is the most accurate reply in a world with tens of thousands of responses to similar questions? The answer? Human verification.
The reality is, there is a massive shortage of accountants. With a retiring demographic, there are not enough new CPAs and accounting majors to replace them at this time. The
With any hype cycle, we get doom and gloom. ChatGPT and generative AI will be no different. In fact, this cycle has already begun. Most tax and accounting professionals are embracing the technology. They are already overworked and exhausted from their client load. It seems a win-win if AI can help them do their jobs more efficiently. However, I'm not sure I want my financial future dependent on a machine. At least not today.
