THE IMPACT OF MACHINE LEARNING ON AUDITING: AUTOMATING FRAUD DETECTION AND COMPLIANCE (A STUDY IN RWANDA)

This study investigates the transformative impact of machine learning (ML) on auditing practices in Rwanda, focusing on fraud detection and compliance automation. Using a mixed-methods approach, quantitative data were collected via surveys of auditing professionals, while qualitative insights were obtained through interviews with stakeholders. The study found that ML adoption increased from 15% in 2020 to 75% in 2024, resulting in a significant improvement in fraud detection accuracy from 65% to 93%. Regression analysis confirmed that a 1% increase in ML adoption correlated with a 1.25% rise in fraud detection accuracy (r=0.96, p<0.01). Compliance issue resolution times dropped from 15 days to 5 days, and audit costs were reduced by 40%. Neural networks emerged as the most effective algorithm, used by 35% of firms. Despite challenges such as technical skill gaps (55%) and integration barriers (50%), the study concludes that ML has revolutionized auditing by enhancing efficiency and accuracy. Recommendations include technical training, financial incentives, tailored algorithms, updated regulations, and cross-sector collaboration.