Summarize this content to 2000 words in 6 paragraphs in Arabic Unlock the Editor’s Digest for freeRoula Khalaf, Editor of the FT, selects her favourite stories in this weekly newsletter.Martin Mulyadi is a professor of accounting and Yunita Anwar is an assistant professor of accounting at Shenandoah University School of Business, Winchester, VirginiaA spate of audit failures has highlighted that traditional auditing approaches may no longer be fit for today’s financial world. But while research suggests artificial intelligence has the potential to significantly improve the efficiency and accuracy of auditing practices, without robust governance frameworks, proper training, ethical AI practices and human oversight, the technology also brings big risks for auditors. AI has the potential to transform auditing practices in a way that is analogous to the impact of the birth of the digital spreadsheet. By automating calculations, that development in 1979 gave accountants more time for decision-making, changing their role in the business world. AI could have a similarly transformative effect. Daniel Davies, managing director at the consultancy Frontline Analysts, describes a future in which AI will unspin the dense data in annual reports, turning static and often unclear documents into dynamic, interactive tools. AI, he says, can sort through huge amounts of financial data, making it possible to spot patterns and inconsistencies. Driven by prospects of enhanced productivity, accountancy firms are investing billions in developing AI tools to handle routine tasks autonomously and are making these integral to their operations. They are also partnering with tech companies such as Nvidia, Microsoft, Google, Oracle and Salesforce to integrate AI into their core services.Accountancy firms are starting to see the benefits. For example, an AI fraud-detection system trialled by EY with 10 UK audit clients flagged suspicious activities at two of those companies. In both cases, the clients later confirmed fraud had occurred, suggesting that AI could dramatically improve audit quality by catching irregularities that traditional methods overlook. As Accenture and Grant Thornton have found, AI also offers efficiency gains. Thomson Reuters beta testers have reportedly halved sample sizes and testing time for certain procedures, while Deloitte believes AI could release its finance agents from thousands of hours’ of work annually, potentially cutting costs by up to 25 per cent. Automating the tedious process of sifting through mountains of data enables auditors to focus on higher risk areas and make complex judgments, allowing accountants to work as strategic advisers using human knowledge to support AI analytics. AI can analyse full data sets, rather than relying on historical sampling, making it easier for auditors to zero in on anomalies. It can also streamline the process of pitching for new business by drawing on databases of past work, which could increase efficiency and profitability. The ‘black box’ nature of AI — the absence of visibility in training data and methods — makes it hard to understand how and why AI technologies arrive at their conclusionsHowever, AI-powered systems also create dilemmas. One academic study identifies challenges such as the bias embedded within AI algorithms and warns auditors to ensure that the decisions they make are fair, accountable and transparent. For example, as an empirical study suggests, the application of large language models to mortgage underwriting has led to higher rejection rates and interest rates for Black borrowers compared to otherwise identical white borrowers. Another concern is the “black box” nature of AI — the absence of visibility in training data and methods. As noted by the Center for Audit Quality, this makes it hard to understand how and why AI technologies arrive at their conclusions. Furthermore, AI technologies are probabilistic — they predict a response rather than retrieving factual data like a search engine. This means that asking the same question multiple times can yield different answers and produce “hallucinations”, or inaccuracies. AI can deliver inconsistent results either because data has been incorrectly structured or because processes are not standardised. One study warns that over-reliance on AI without adequate human scrutiny — often referred to as “human-in-the-loop” — the technology could undermine auditors’ ability to identify nuanced irregularities or fraud. Deloitte and KPMG have voiced similar concerns, arguing that if AI is trained on past cases of fraud, it may not detect new forms of fraud designed to circumvent existing safeguards. Companies need to balance automation and human judgment, equipping employees to evaluate AI outputs and understand the limitationsAs a result, firms must tread carefully when incorporating AI into accounting and reporting. Robust governance, ethical oversight and good data management are essential. Meanwhile, companies need to balance automation and human judgment, which means equipping employees to evaluate AI outputs and understand the technology’s limitations. Implementing AI also requires a holistic strategy driven by senior leadership. All this comes with high costs and an uncertain return on investment. Yet companies that are slow to adopt AI in auditing also face risks such as lower levels of efficiency and audit quality, according to accountancy training firm Mercia Group. Meanwhile, firms may encounter difficulties hiring top talent, since employees increasingly prefer to work with new technologies. While AI offers the potential for increased productivity and efficiency, incorporating AI in accounting and auditing presents many challenges. Accounting and auditing professionals must weigh the transformative power of AI-powered systems against the responsibilities and risks that come with using them.
