Artificial Intelligence Adoption and the Financial Performance of Commercial Banks in Kenya

Commercial banks in Kenya have invested heavily in artificial intelligence (AI), yet sector profitability has weakened, with average return on assets (ROA) falling from 3.3 percent in 2021 to 2.6 percent in 2024. This research investigated the effect of AI adoption on the financial performance of commercial banks in Kenya. Specifically, it examined the effects of AI-enabled customer service, machine learning credit scoring, robotic process automation and AI-driven fraud detection on financial performance, measured by ROA, as well as the mediating effect of operational efficiency and the moderating effect of bank size. A census of all 38 commercial banks licensed by the Central Bank of Kenya was undertaken, using data from audited annual reports and Central Bank of Kenya supervisory returns. Data were analysed using random effects panel regression with cluster-robust standard errors, the causal steps approach supplemented by the Sobel test, and moderated regression. AI-enabled customer service had a positive and significant effect on ROA (β = 0.927, p = 0.003). Machine learning credit scoring (β = −0.046, p = 0.841), robotic process automation (β = −0.450, p = 0.098) and AI-driven fraud detection (β = 0.137, p = 0.459) had no significant effects. AI-enabled customer service lowered the cost-to-income ratio (β = −0.194, p < 0.001), but operational efficiency did not significantly mediate its relationship with ROA (Sobel z = 1.287, p = 0.198). The interaction between AI adoption and bank size was not significant (β = 0.154, p = 0.589). The study concluded that the effect of AI adoption on bank performance is application-specific. Only customer-facing AI translated into higher profitability, while the credit scoring, process automation and fraud detection applications did not. Operational efficiency did not mediate, and bank size did not moderate, the connection between AI adoption and performance.