Search

The press and pundits have extensively discussed the dangers of artificial general intelligence (AGI) and the various ethical and functional effects of current emerging technologies. This paper attempts to focus on AI as applied to accounting and auditing and its realities. It shows studies using machine learning and natural language processing, including generative efforts, and robotic and extended process automation. It concludes that a very promising set of applications is emerging, but there is little immediate danger of being replaced by machines to most in the profession, although the impacts are going to be substantive. These applications primarily utilize individual or combined AI functions, and the realization of true intelligence remains a distant goal. JEL Classifications: M41; M42; C45. [ABSTRACT FROM AUTHOR]

As environmental, social, and governance (ESG) reporting has become a mainstream channel for companies to communicate their commitment to sustainability issues, the need for reliable and transparent ESG reports is increasing. However, research on ESG assurance is still in its early stages. ESG assurance poses more challenges than traditional financial auditing due to the diverse subjects and types of information in ESG reports. This paper proposes using artificial intelligence (AI) technologies and exogenous data as solutions. It discusses how AI can enhance the efficiency and effectiveness of ESG assurance by assessing vast and extensive data. This paper also explores AI's application throughout the general ESG assurance process and contributes to the discussion on providing high-quality ESG assurance services. Additionally, it provides practical implications for auditors, regulators, and stakeholders. [ABSTRACT FROM AUTHOR]

The emergence of large language models (LLMs) presents significant opportunities in accounting, including optimizing current processes, extracting new information, and updating accounting measurements. However, factors such as skill gaps, perceived complexity of integration, and cost constraints have limited the implementation of LLMs in accounting. This study provides an overview of mainstream LLM utilization methods, including user interface and application programming interface, and introduces a novel approach via robotic process automation integration. The advantages and limitations of each method are discussed, accompanied by a current analysis of the time, labor, and monetary costs involved in using LLMs for accounting tasks. To facilitate practical applications, three off-the-shelf examples are provided. This study contributes to the literature and practice by summarizing and comparing LLM implementation methods, responding to the challenges raised by researchers and stakeholders, and bridging the gap between technology innovation and its practical application in accounting. JEL Classifications: M41; O14; O33; D61. [ABSTRACT FROM AUTHOR]

This study constructs a measure of audit quality that captures the effect of potential factors that are generally unobservable to people outside of the audit firm or client company. Using machine learning and a wide range of data describing audit firm characteristics, audit partners, and public companies in China, this paper constructs the "surprise score," a new measure of audit quality, calculated as the difference between the predicted probability and the actual value of an audit quality-related event (i.e., the existence of material misstatements, audit adjustments, and nonclean audit opinions). The effectiveness of the surprise score is validated by testing the association between the surprise score and penalties or audit firm changes. The proposed approach is applied to U.S. data to generalize its application. The surprise score adds value to existing audit quality measures and can help regulators to make better-informed decisions about audit quality. Data Availability: Except for the data privately provided by CICPA and MFC, other datasets are available from the public sources cited in the text. JEL Classifications: M41; M42. [ABSTRACT FROM AUTHOR]

Recent studies and new standards suggest that auditors can use information from expanded external sources to verify their clients' financial information. We propose advanced substantive analytical procedures with external nonfinancial measures derived from weather information to test whether it helps detect misstatements. Using computational simulations and daily store-level sales data, we test whether the proposed procedures with weather indicators outperform the procedures without such indicators in substantive analytical procedures to identify overstated daily store sales. For the multilocation retail firm examined in this study, we find that the models with one or more weather indicators perform better at detecting misstatements than the models without them. When the reliability of relevant internal information is in question, the usefulness of weather indicators is apparent. Overall, our results provide evidence suggesting the potential value of external nonfinancial measures in auditing. [ABSTRACT FROM AUTHOR]

Auditors traditionally use sampling techniques to examine general ledger (GL) data, which suffer from sampling risks. Hence, recent research proposes full-population testing techniques, such as suspicion scoring, which rely on auditors' judgment to recognize possible risk factors and develop corresponding risk filters to identify abnormal transactions. Thus, when auditors miss potential problems, the related transactions are not likely to be identified. This paper uses unsupervised outlier detection methods, which require no prior knowledge about outliers in a dataset, to identify outliers in GL data and tests whether auditors can gain new insights from those identified outliers. A framework called the Multilevel Outlier Detection Framework (MODF) is proposed to identify outliers at the transaction level, account level, and combination-by-variable level. Experiments with one real and one synthetic GL dataset demonstrate that the MODF can help auditors to gain new insights about GL data. Data Availability: The real dataset used in the experiment is not publicly available due to privacy policies. JEL Classifications: M410, M42. [ABSTRACT FROM AUTHOR]