AI tools are linked to faster, more accurate external audits in Taraba State banks, with audit automation producing the largest reported efficiency gains; however, the result relies on self-reported survey data from local audit staff rather than objective or causal evidence.

Main Finding

AI adoption in Deposit Money Banks in Taraba State (2021–2025) — measured as AI-based audit automation, machine‑learning analytics, and AI fraud‑detection systems — has a positive and statistically significant effect on external auditing efficiency. AI‑based audit automation exhibited the largest effect among the three. Overall, integrating AI tools improved audit timeliness, accuracy, workload reduction, and fraud detection.

Key Points

• Independent variables studied: AI-based audit automation, machine learning analytics, AI fraud detection systems.
• Dependent/efficiency outcomes: audit timeliness, audit accuracy, reduction in manual workload, fraud detection capability.
• Empirical result: all three AI dimensions positively and significantly influence external auditing efficiency; automation had the strongest impact.
• Reported benefits: shorter audit cycles, fewer manual procedures, better identification of anomalies/fraud, improved reliability of audit evidence.
• Main barriers noted: limited digital skills among auditors, inadequate IT infrastructure, implementation costs, data quality concerns, potential over‑reliance on AI outputs, and regulatory ambiguity in Nigeria.
• Policy/practical recommendations: invest in AI tools, upskill auditors’ digital competencies, strengthen regulatory guidance/standards for AI in audit, and improve banking IT infrastructure.

Data & Methods

• Context and period: Deposit Money Banks in Taraba State, Nigeria; 2021–2025.
• Respondents/source: primary data from external auditors and bank audit‑related staff (internal control officers, compliance officers, finance officers, etc.).
• Data collection: structured questionnaires.
• Analysis methods: descriptive statistics, correlation analysis, and multiple regression analysis to test the relationships between AI adoption dimensions and audit efficiency.
• Limitations implied by design: cross‑sectional/survey data from a single state; no experimental or administrative causal identification reported; potential measurement and self‑reporting biases.

Implications for AI Economics

• Productivity and labor composition: Results indicate that AI raises auditor productivity (faster audits, lower manual workload) and shifts auditor tasks from repetitive processing toward judgment, risk assessment, and oversight of AI outputs. This supports theories of task augmentation rather than outright substitution in professional services.
• Cost structure and returns to investment: Positive efficiency effects imply reduced audit time and operating costs per engagement, suggesting potentially high private returns to banks and audit firms that invest in AI. However, upfront implementation costs and recurrent data/IT investments temper net returns—especially for smaller or regional banks.
• Human capital and wages: Increased demand for digital competencies implies rising returns to training and technical audit skills; wage and hiring patterns may shift toward more data‑savvy auditors. Public policy or firms may need to subsidize upskilling to realize AI gains.
• Adoption heterogeneity and diffusion: Infrastructure and skill gaps (highlighted in Taraba State) imply uneven diffusion of audit AI across regions and bank sizes, producing divergent productivity and audit‑quality outcomes. This heterogeneity matters for market competition and regulatory oversight.
• Information asymmetries and financial stability: Improved fraud detection and audit accuracy reduce informational frictions between banks and stakeholders, which may lower monitoring costs and market risk premia; broader adoption could generate positive externalities for financial sector transparency.
• Regulation, standards, and governance: The findings point to a need for regulatory guidance (audit standards for AI, model governance, explainability, data privacy) to manage risks (algorithmic bias, over‑reliance, false positives) and to standardize practices—this will affect compliance costs and shape market incentives.
• Research and policy priorities: Economic evaluations should quantify net social returns (costs of implementation plus governance vs. efficiency and fraud‑loss reductions), study labor reallocation effects, and model diffusion under varying infrastructure and regulatory scenarios. Replication beyond a single state is needed to generalize welfare and market‑level conclusions.

Caveats - State‑level, survey‑based study: external validity beyond Taraba State and the Nigerian banking context is limited.
- No causal identification strategy reported (e.g., instrument, difference‑in‑differences), so results should be read as associations consistent with positive AI effects rather than definitive causal estimates.
- Detailed numeric estimates (coefficients, p‑values, effect sizes) were not provided in the text excerpt; for policy modelling, a follow‑up with the full paper or data would be necessary.

Suggested next steps for economic research - Conduct cost‑benefit and ROI analyses of AI adoption in audits across bank sizes and regions.
- Use panel or quasi‑experimental designs to estimate causal impacts on audit time, costs, and fraud incidence.
- Model labor market adjustments (retraining needs, wage dynamics) in the auditing profession under AI diffusion.
- Assess systemic effects on financial stability and market discipline from improved audit quality.