During an executive leadership meeting earlier this year, we allocated a portion of our agenda to delve into the capabilities of ChatGPT. Our AI team leaders showcased a wide array of functionalities, encompassing content generation for marketing, code generation, customer support interactions, and even tackling a highly intricate accounting query. Towards the conclusion of the presentation, our accomplished global corporate controller, who is young and high-achieving, exclaimed, “I might be replaced by AI!” In response, another one of our leaders, renowned for his logical thinking and dry humor, quipped, “No, you’ll be succeeded by someone who harnesses AI more effectively than you do.” This leads us to the question at hand: will ChatGPT and other generative AI technologies supplant entire industries like accounting, or will they, in the Sage spirit, “enhance human work”?
To clarify, the accounting industry in Australia is already heavily integrating AI with remarkable effectiveness. AI systems capable of reading and categorizing data from business documents are becoming commonplace. Accountants are employing AI for automating financial modeling and forecasting. Furthermore, as a byproduct of this automation, accountants are relying on AI to detect unusual transactions, pinpoint accounting errors, unveil shifts in business performance, and even expose fraudulent activities.
So, what sets generative AI apart? Let me provide a brief explanation. Generative AI refers to a category of artificial intelligence that can generate fresh content or outcomes, often based on patterns or structures it has learned from existing data. This encompasses generating text, images, music, and intricate designs. GPT, the foundation for ChatGPT, falls under the category of generative AI known as Large Language Models (LLMs). LLMs are machine learning models designed to comprehend, interpret, and generate text that closely resembles human language. These models acquire language patterns and structures from vast amounts of textual data, allowing them to complete prompts, answer queries, or even create entire articles in a remarkably human-like manner.
Many of you have likely already experienced a revelatory moment with ChatGPT or another new chatbot, where you realized AI has taken a significant leap forward. For me, it was a simple interaction that left a profound impact. I shared a dad joke with ChatGPT: “A skeleton walks into a bar and says to the bartender, ‘I’ll have a beer and a mop.'” While many humans don’t grasp this joke immediately, ChatGPT swiftly responded, “Haha, that’s a good one! The joke plays on the fact that a skeleton, being made of bones with no organs or tissues, would not be able to consume the beer like a living person. Instead, the beer would simply flow through its bones and make a mess on the floor, hence the need for a mop.” After picking my jaw up off the floor, I realized that OpenAI, the creators of ChatGPT, had overcome two significant AI limitations: understanding the world and the ability to reason. Whether these abilities are genuine or simulated is immaterial – AI can now perform complex tasks with minimal guidance.
Research conducted by OpenAI in partnership with the University of Pennsylvania indicates that approximately 80% of the workforce will see at least 10% of their work tasks affected, with the remaining 20% experiencing at least a 50% impact. Unsurprisingly, jobs in accounting, auditing, and tax preparation are expected to be among the most affected. The research generated numerous headlines, with my favorite being somewhat cheeky: “Stonemasons Should be Fine in the Brave New World.” Evidently, journalists, who also face significant disruption, employ sarcasm as a natural defense mechanism.
What does all of this signify for the Australian accounting industry? Let’s begin with a significant caveat: generative AI is advancing rapidly and will be profoundly disruptive. I believe that generative AI is as transformative as the Internet or the smartphone but is racing through the adoption curve faster than any technology we’ve ever witnessed. ChatGPT reached one million users in five days. It reached 100 million users in two months. In just six months, its capabilities, as measured by performance on the Bar exam, improved from being in the bottom 10% to reaching the top 10%. Predicting the outcome is impossible, but I can outline a few probable scenarios.
AI will evolve from automating tasks to automating entire workflows. As mentioned earlier, we can currently use AI to extract data from an invoice, categorize expenses, and identify issues like overbilling. Presently, this AI operates within well-defined workflows where pre-configured rules dictate the actions. With LLMs, we will be able to delegate the orchestration to “AI Agents” capable of operating autonomously. These agents could manage and execute workflows with instructions akin to those given to an accountant: “ensure the cash balance remains above X; prioritize early payments to vendors of type Y; alert me to any unusually large bills.”
Digital Assistants will revolutionize how we interact with technology. Current business applications guide user interactions through predefined menus, meticulously designed data entry forms, sorting and filtering lists, and preformatted reports or dashboards for data analysis. These user interfaces are both limiting and challenging to learn. LLMs hold the promise of replacing all these with simple, human-language interactions. Imagine working within a spreadsheet and instructing a digital assistant to gather the accounting data needed for analysis. Picture reviewing and approving purchase orders directly within collaboration platforms like Teams. Tools such as Excel and Teams are familiar and tailored to specific tasks, enabling more user-friendly, self-service capabilities that reduce reliance on the accounting team. Furthermore, envision LLMs integrated with enterprise systems, allowing employees within the business to obtain answers or conduct data analysis via a straightforward chat interface. As chat interfaces become the norm, employees will no longer tolerate the limitations of today’s rudimentary tools.
AI will handle routine interactions with customers and vendors. The majority of emails received by accounting teams fall into a limited set of repetitive interactions. For instance, customers inquire about tax information, request their latest statement, or provide payment details; meanwhile, vendors send invoices, inquire about payment dates, and supply compliance information. LLMs excel at comprehending the intent of these emails, determining the necessary actions (e.g., sending an invoice for processing or retrieving an invoice from the accounting system for a customer), and generating response emails when required.
AI will serve as an expert resource. Accountants will gain access to an expert resource for understanding and applying accounting standards. What is the correct accounting treatment for a complex sales contract? What does tax law dictate regarding capital asset write-offs? LLMs are also adept data analysts. Provide an LLM with access to balance sheet data for the past 24 months and ask it to identify the most significant trends. Provide detailed ledger data, and it can forecast future cash flow. More broadly, AI will enhance the productivity of every knowledge worker. A few paragraphs ago, I struggled to craft a concise, comprehensible definition of generative AI and LLMs, so I turned to ChatGPT for assistance. While its suggestion wasn’t flawless, it provided a starting point and unblocked the writing process.
Of course, LLMs come with well-documented risks. They are known to generate responses that are articulate, well-reasoned, and entirely fabricated, a phenomenon referred to as “hallucination.” They can be manipulated into behaving inappropriately. Concerns about data privacy abound. The use of LLMs has raised numerous legal questions, both regarding the content used to train the models and the ownership of the content they generate. Technology is outpacing regulation, and the situation may necessitate a moratorium on their use. However, the genie is already out of the bottle. We need the accounting industry in Australia to offer guidance to businesses on how to harness AI safely and effectively.
I traveled to New York City earlier this year to meet with my colleagues at the AICPA (American Institute of Certified Public Accountants). I wanted to gauge their reactions and detect any signs of apprehension. What I witnessed was quite the opposite – where most perceive a threat, they discern opportunity. We identified two overarching opportunities with generative AI: 1) addressing the industry’s labor shortage and 2) enabling accountants to provide greater strategic value. As we delved deeper into the risks and challenges posed by generative AI, we uncovered a third opportunity that I believe surpasses the first two. As we delegate more and more tasks to technology, especially unpredictable technology, the value of human accountability will rise.
Addressing the industry’s labor shortage. The accounting industry in Australia is well aware of its labor shortage. The number of employed accountants and auditors declined by 17% between 2019 and 2021, according to Bloomberg Tax. Enrollment in accounting courses at universities decreased by four percent between 2016 and 2019. A significant majority of hiring managers for accounting and financial positions in public companies, 82.4% to be precise, consider recruitment a significant challenge, as per Deloitte & Touche LLP. While we can debate how to attract more individuals to accounting careers, a more effective solution may be to boost accountants’ productivity. AI is particularly adept at automating repetitive tasks such as data entry and reconciliation. Sage customers using our AI-powered invoice processing have reported a doubling or even tripling of productivity, creating more capacity within their accounting teams. This automation also has a positive side effect – automating low-value, repetitive work may make an accounting career more appealing.
Enabling accountants to provide greater strategic value. There is no negotiation when it comes to compliance-related work in the list of tasks assigned to accountants. Audit cannot be postponed to make room for evaluating a potential acquisition. Tax filing deadlines are non-negotiable. Time and resources allocated to strategic work can only follow the mandatory tasks. Automating these non-negotiable tasks naturally frees up time for more strategically valuable work. Equip accountants with LLMs, and not only will they have more time for such work, but they will also possess an invaluable, hyper-efficient research assistant capable of analyzing vast amounts of information and structuring coherent, professional presentations of the results.
The value of human accountability. At Sage, we have a guiding principle – we should match our investments in automation with equivalent investments in trust. Human users will only entrust their work to technology if they have confidence in the technology’s ability to perform the task safely and competently. While transitioning work to technology is possible, accountability will always rest with humans. Audit statements are signed by humans, not AI.
The accounting industry’s most significant contribution to society is instilling confidence in our markets. There are various ways to measure the value of that confidence. For example, the global big four accounting firms generated $271 billion in audit fees in 2021. I believe AI will substantially increase this value exponentially, partly due to the increased risk introduced by AI, but more importantly because a more efficient and productive accounting industry will lead to more efficient and productive markets.
Returning to our question of the day, I am quite optimistic. With generative AI, we have the opportunity not only to enhance human work but also to elevate the entire accounting industry in Australia. Yes, we should proceed cautiously, but we should also proceed with determination and urgency. The winners will be those who leverage AI more effectively than anyone else.
Original content source: Sage Advocacy/ diginomica.com – Photo by Possessed Photography on Unsplash