Impact of AI on the Accounting Profession

Some years ago I had the opportunity to work for a blockchain company. While with the company, I authored an article titled, “When Blockchain Theory meets Accounting Reality.” The focus of the article had to do with the potential of blockchain to upend the accounting industry. Sound familiar? Even Deloitte considered the topic important enough to write a whitepaper about it titled, “Blockchain Technology – A game-changer in accounting?” At the time, I even bought into the hype and thought blockchain was going to be the biggest innovation since creation of the internet.

So… what happened?

Needless to say, blockchain has not turned out to be the disruptive force as expected, primarily due to significant challenges including scalability limitations, a lack of regulatory clarity, poor user experience, high implementation costs, and interoperability issues. These practical barriers have slowed its mainstream adoption across various industries beyond its primary use in cryptocurrencies. Even though blockchain has not been the disruptive force many thought it would be, it has shown its value in areas like transaction tracking (for auditing), smart contracts and primarily acting as a sustaining innovation that improves existing systems discreetly rather than disrupting entire industries.

So, now AI has replaced blockchain as the “big name on the block.” Just google it and you can see that it is being widely promoted as the new transformative technology. In fact, prominent figures such as Google’s Sundar Pichai and Apple’s Tim Cook, along with Microsoft reports, compare AI’s impact to that of electricity and the internet, predicting it will significantly reshape industries and everyday life.

So…let’s begin by defining what AI is.

What is AI?

“Artificial intelligence (AI) is a set of technologies that empowers computers to learn, reason, and perform a variety of advanced tasks in ways that used to require human intelligence, such as understanding language, analyzing data, and even providing helpful suggestions. It’s a transformational technology that can bring meaningful and positive change to people and societies and the world.” (Google Cloud)

Diving deeper into the definition, AI is a field in computer science dedicated to creating intelligent machines, especially computers that can simulate human intelligence processes. AI uses algorithms, which are sets of step by-step instructions that a computer can execute to perform a task. Some AI applications are able to learn from data and self-correct, according to the instructions given.

To help contextualize the use of AI, researchers categorized it in two primary ways: by its capabilities and by its functional subsets.

Capabilities

These levels of AI are categorized as either weak AI, also called narrow AI, or strong AI, also called artificial general intelligence (AGI). They refer to the evolutionary stages of AI, from systems currently being used to theoretical future ones.

• Weak AI or narrow AI is an AI system that can simulate human cognitive functions but although it appears to think, it is not actually conscious. A weak AI system is designed to perform a specific task, “trained” to act on the rules programmed into it, and it cannot go beyond those rules.

  • Voice recognition software like Apple’s Siri and Amazon Alexa are examples of weak AI. It has access to the whole Internet as a database and is able to hold a conversation in a narrow, predefined manner; but if the conversation turns to things it is not programmed to respond to, it presents inaccurate results.

  • Industrial robots and robotic process automation (RPA) are other examples of weak AI. Robots can perform complicated actions, but they can perform only in situations they have been programmed for. Outside of those situations, they have no way to determine what to do.

• Strong AI or AGI is equal to human intelligence and exists only in theory (think Data from “Star Trek” or Isaac from Seth MacFarlane’s series “The Orville”). A strong AI system would be able to reason, make judgments, learn, plan, solve problems, communicate, create, and build its own knowledge base, and program itself.

Functional Subsets

These are the specific technologies and fields that fall under the general umbrella of AI, or as I refer to as the AI universe. This section addresses the identification and differentiation of key subsets within the AI universe which are most pertinent to the accounting profession.

Accountants should get to know these subsets for three main reasons:

1. Relevance to the Accounting Profession: Accountants increasingly encounter AI in their work environments. Knowing the different subsets of AI helps them better understand the tools and technologies impacting their daily tasks.

2. Enhanced Decision-Making: Familiarity with AI subsets enables accountants to make informed decisions when selecting or recommending software and analytical solutions. It also equips them to recognize which AI-driven processes can improve accuracy and efficiency.

3. Adapting to Technological Change: As AI continues to evolve, accountants who are aware of its subsets can adapt more readily to changes. This adaptability is crucial for staying competitive and relevant in the industry.

Subsets of the AI Universe 

Within the AI universe, four main subsets are particularly relevant to accounting: Machine Learning (ML), Deep Learning (DL), Natural Language Processing (NLP), and Generative AI (GenAI).

Understanding how these subsets relate to the broader AI universe is similar to the progression from learning basic accounting principles (dual-entry accounting, transaction classification, financial reporting) to mastering advanced topics (managerial accounting, tax, audit, forensic auditing) and software (QuickBooks, Xero, Oracle’s NetSuite, SAP).

Both human and artificial "brains" process information, learn, solve problems, and recognize patterns, though their underlying mechanisms are quite different.

AI functions as the "brain" or capability set (like machine learning algorithms) that, when implemented in computers, enables them to perform intelligent tasks. The evolution of AI mirrors the typical learning curve, starting with simple instructions and growing in complexity as user needs expand. Both accountants and AI systems continuously learn, adapt, and improve.

These AI subsets often overlap in technology and application. For instance, Machine Learning (ML) underpins many advancements and connects with Natural Language Processing (NLP) and computer vision. Generally, when people refer to AI today, they’re speaking of Machine Learning. For this particular article, we take a closer look at Machine Learning and how it helps accountants do their work better and more efficient.

Machine Learning

The diagram above show us that “Machine learning (ML) is a subset of AI that enables a system to autonomously learn and improve using neural networks and deep learning, without being explicitly programmed, by feeding it large amounts of data. ML allows computer systems to continuously adjust and enhance themselves as they accrue more ‘experiences.’ Thus, the performance of these systems can be improved by providing larger and more varied datasets to be processed.” (Google Cloud)

For accountants, a high-level understanding of ML can build confidence in adopting new tools, instead of relying solely on step-by-step instructions. As the diagram below shows, Machine Learning (ML) fundamentally involves systems learning from data to find patterns, using that knowledge to predict outcomes or make decisions, and then improving their accuracy and performance over time without explicit reprogramming, essentially enabling computers to learn from experience.

Machine Learning in Accounting

Machine learning (ML) has been an instrumental tool for accountants by automating tedious tasks, improving accuracy, detecting anomalies and fraud, enhancing forecasting, and shifting accountants’ focus from transactional work to strategic analysis. Some of the key applications of ML in accounting include:

• Task automation (data entry, invoice processing, bank reconciliations).

• Fraud detection and risk assessment.

• Data analysis and insights for better decision-making.

• Financial forecasting and planning.

• Continuous auditing and compliance monitoring.

• Tax preparation and compliance assistance. 

Illustration: Categorizing Bank Transactions

An example of machine learning (ML) in accounting is using ML to automatically categorize bank transactions based on historical patterns and data.

This process is found in software programs like QuickBooks online and Xero. These programs significantly reduce the need for manual data entry and improves efficiency.

The process of automating bank transactions includes:

1. Training phase:

The ML algorithm is provided with a large dataset of a company's past financial transactions (e.g., two years of bank data) that have already been manually categorized by a human accountant.

The algorithm "learns" the association between transaction details (vendor name, amount, date, description) and the correct general ledger expense codes, e.g., associating "Starbucks" with "Business Expense" and not with "Office Supplies".

2. Application phase:

When a new, uncategorized transaction is imported into the accounting software, the ML model analyzes its details. Based on the patterns it learned, the system automatically suggests the most probable category.

3. Human feedback & Continuous Improvement:

As an accountant you still need to review the ML system's suggestion. You can accept the suggestion or manually override it if the context is different, e.g., the Starbucks charge was for a client meeting and not office supplies.

This human feedback is used by the algorithm to continuously refine its model, becoming more accurate and reliable over time without explicit reprogramming.

Other Illustrations of Machine Learning in Accounting:

• Automated Bookkeeping and Data Entry: ML algorithms use OCR to extract, categorize, and post financial data from documents, learning from corrections to improve over time.

• Fraud Detection and Anomaly Capture: ML analyzes past transactions to detect anomalies or fraud more efficiently than manual or rule-based audit processes.

• Predictive Analytics and Financial Forecasting: ML enhances forecasting of trends, cash flow, and expenses by processing large datasets, aiding proactive business decisions.

• Enhanced Auditing: ML reviews all transactions, flags high-risk cases for auditors, and streamlines the audit process.

Final Word  

Machine learning (ML) is not replacing accountants but it is transforming their roles. By automating menial, repetitive data processing tasks, ML frees up time for accountants to focus on more strategic, high-value activities like financial analysis, advisory services, complex problem-solving, and client relationship management.

For my next article, we progress further down the AI universe and discuss Natural Language Processing (NLP) and Generative AI (GenAI). We would be remiss if we didn't bring Agentic AI into the discussion. In 2025, Agentic AI went from experimental automation to an autonomous "digital teammate" in accounting.

About Carl Burch

Carl Burch holds an MBA, CMA, CIA, and FCCA and is founder of BURCH Business Services (BBS) located in Boston, MA. For more information on BBS, visit www.burchbusinesservices.com