Academy of Accounting and Financial Studies Journal Volume 25, Issue 6, 2021 1 1528-2635-25-6-908 Citation Information: Fallatah, R. (2021). Literature review – machine learning in accounting and assurance. Academy of Accounting and Financial Studies Journal, 25(6), 1-14. LITERATURE REVIEW – MACHINE LEARNING IN ACCOUNTING AND ASSURANCE Rasha Fallatah, Taif University ABSTRACT The purpose of this literature review is to gain an understanding of the existing research or literature relevant to the impact of Machine Learning on the profession of accounting and assurance. It provides a comprehensive overview of knowledge about this area of study in the form of a written report. Specifically, the study describes the impacts of ML on accounting and assurance in the following manner; by identifying the emerging trends in these fields, by identifying the impact of ML on the efficiency and effectiveness of these professions, and the inconsistencies and gaps in the research. The details of this study are helpful for the professionals in accounting and auditing as it enhances the readers understanding of the topic at large. Keywords: Artificial Intelligence, Machine learning, Blockchain, Digitization. INTRODUCTION Arthur Samuel first introduced machine learning (ML) in 1959. It is a subfield of Artificial Intelligence that aims to enable computer programs to automatically learn and develop itself by using large amounts of data to provide valuable insights and future predictions (Kaur, 2020). Machine learning consists of neural networks that are like the skills of the human brain. It is developed to mimic human brainstorming patterns. Therefore, after analyzing the sufficient data, it starts interpreting them and connects them to take necessary actions on its own (Helm et al., 2020). Unlike the traditional computer programs designed to give distinct instructions to the computers for solving a defined problem with both the possibilities of certainty and uncertainty (Rice, 2014). ML enables software applications to become more accurate by using historical data as input and predicting outcomes that are not explicitly programmed to do so (Michie et al., 1994). The fundamental categories of algorithms in ML are supervised, unsupervised, and semi-supervised algorithms (Ayodele, 2010). These algorithms are trained to change and improve themselves, analyze anomalies, remove errors without any human intervention and mitigate the chances of occurring again (Bonaccorso, 2017). They are also used to improve processing speed, review source documents, and find similar patterns from huge or complex data (Mohammed et al., 2016). Machine learning holds a significant attraction for business world in these contemporary times. The broad range of facilities that it offers and the various applications on business data that it has, allows the organizations to easily cope with the dynamic environmental conditions in diversified industrial sectors (Apte, 2010). ML is beneficial in accomplishing complex business tasks with great accuracy instead of humans who cannot process huge quantum of data and produce accurate conclusions (Finlay, 2017). Similarly, the integration of multiple processing units results in a high processing speed and decreases the element of human biases (Canhoto & Clear, 2020). Today, extensive research is being conducted on the impacts of machine learning on the professions of accounting and assurance. Its vast application on various tasks such as assessing business risks, analyzing business transactions or activities, and reviewing source