Since the launch of ChatGPT in November 2022, Generative Artificial Intelligence (Gen-AI) has been at the forefront of technological discussions, revolutionising industries with its rapid advancements. It is constantly evolving, with new tools emerging every month. The impact has been so significant in terms of intensity and agility that it now seems impossible for anyone to function as they did before this AI boom.

According to McKinsey, AI adoption has more than doubled over the past five years, and investment in AI is increasing rapidly, contributing $4.4 trillion to the global economy annually. Organisations are eagerly incorporating Gen-AI into their operations: McKinsey’s Global Survey (2024) reveals that 65 percent of organisations are regularly using Gen-AI, nearly double the percentage from the previous survey conducted just ten months ago. Moreover, according to another survey conducted by KPMG, nearly three-quarters (72 percent) of businesses are already exploiting AI, a figure expected to reach almost 99 percent in the next three years.

What is a Financial Audit (hereinafter referred to as “Audit”)? For the readers of this piece, let us outline an audit primer in simple terms: Most businesses are run by managers rather than the owners themselves, and naturally, shareholders with stakes in the company are keen to know the financial calibre of their investments. To satisfy this need, the management of the company prepares—mostly—annual financial statements (FS). Though these statements are meant for various stakeholders, they have the most impact on shareholders, as the numbers stated directly affect the valuation of their investments.

Now the question arises: how can the owners be sure about the credibility of the data presented to them? This is where the auditor comes in. The auditor performs audit procedures and provides an opinion—called an audit opinion—about the conformity of the FS prepared by management with International Financial Reporting Standards (IFRS) or any other applicable financial reporting framework, helping shareholders make informed financial decisions.

Businesses in every sector have strived to incorporate AI into their operations to achieve an efficient and dynamic modus operandi. Accountancy firms are no exception. In the words of Matthew Campbell, Chief Technology Officer of KPMG UK: “There’s so much more to AI than automating simple tasks like analysing data. AI can support all areas of the audit. From helping auditors to make more insightful judgments to providing a more robust challenge to management too.”

Deloitte, PwC, EY, and KPMG are spearheading the integration of AI into their auditing processes, leveraging technology to enhance accuracy and efficiency. In their pursuit of enhanced auditing capabilities, Gen-AI/AI is currently being used mostly for automating repetitive tasks (such as data entry and processing, report generation, and reconciliation), enhancing accuracy, and identifying errors in financial data processing (such as error identification and compliance checking). Additionally, AI is used for predictive analysis to identify potential risks based on historical data and current trends, as well as to assist in financial forecasting by analysing past data and predicting future trends. AI also serves as a decision support mechanism and interactive assistant, providing real-time support and answering queries relating to audit processes and procedures.

That said, is there any possibility of a full-fledged audit report/opinion being produced by AI alone with (almost) no human intervention? The answer to this is “No”, with the caveat “as of now”.

Why is this the case? To understand, one needs to delve into the inner workings of Gen-AI itself: Gen-AI is an incredible AI model that mimics human intelligence and is capable of creating new unstructured content such as text, images, and music, unlike its predecessors, which are more analytical and mathematical in nature. Yet, as many of us have experienced, it can “hallucinate” and provide nonsensical answers, sometimes even performing basic arithmetic operations incorrectly. This occurs because it predicts what a human might enjoy or find useful rather than performing sophisticated calculations. As a result, it can produce radically erratic results. For instance, the new AI Overview feature rolled out by Google has stirred controversy by suggesting glue in place of cheese sliding off pizza or advising a person feeling depressed to jump off the Golden Gate Bridge.

Furthermore, a study conducted at the London School of Economics examined how AI affects the financial system. It was noted that, in basic scenarios such as chess, where the pieces on the board and the rules are well-defined, AI easily outperforms humans. However, its advantage diminishes as complexity increases. In unexpected situations, humans can draw on a wide range of knowledge—from economics and history to ethics and philosophy—to make better-informed decisions. This, the study concluded, is where humans currently surpass AI.

Will the situation remain like this? Of course not. Who would have predicted the prevalent usage of GPT just about a decade ago? This intervention has been significant; the continual release of new tools every now and then marks one of the most dynamic periods in human history. The speed at which these changes are being adopted is unprecedented.

So, how long will it take for AI to become sophisticated enough for shareholders to rely on it? Well, that depends on when we can overcome the current limitations of AI development. A term often used in this context is “Artificial General Intelligence” (AGI). AGI represents a form of Gen-AI with all its current limitations mitigated. It is a theoretical construct that would not suffer from the issues traditional Gen-AI faces, such as poor contextual and analytical skills, creating factually incorrect essays, and providing problematic and counterintuitive answers.

Bernard Marr, a world-renowned futurist, board advisor, and author of “Generative AI in Practice: 100+ Amazing Ways Generative Artificial Intelligence is Changing Business and Society”, while writing for Forbes, states that: “In stark contrast to the specific applications of current AI systems, AGI represents a theoretical pinnacle of this technology. Unlike specialised AI, AGI would be capable of understanding and reasoning across a broad range of tasks. It would not only replicate or predict human behaviour but also embody the ability to learn and reason across diverse scenarios, from creative endeavours to complex problem-solving. To do that, it would require not just intelligence but also emotional and contextual awareness.”

Although we are not there yet, the rapid pace of AI development suggests that this future may be closer than we think. Experts estimate that we could achieve the once-distant dream of AGI by 2030 at the latest, and possibly as early as 2027.

As AGI becomes a reality, its impact on audit opinions could be transformative. Unlike current AI, which automates repetitive tasks and assists with predictive analysis, AGI would bring a level of intelligence and contextual understanding previously reserved for human auditors. AGI’s ability to process vast amounts of data, learn from diverse scenarios, and apply contextual awareness would enhance the accuracy and reliability of audits. This could fundamentally reshape how audit opinions are formed and trusted by shareholders, potentially reaching a level of trust where human intervention in auditing becomes minimal.

Furqan Ali & Abdullah Ahmed
Furqan Ali is a Policy Fellow at Learner’s Republic and presently serves as an advisory associate at a firm based in Peshawar. Abdullah Ahmed is a Policy Fellow at Learner’s Republic and a Data Science & AI trainee at Atomcamp.