Artificial Intelligence Meets Real Finance: Innovation, Risk, and Regulation

Artificial intelligence is reshaping financial services by improving credit scoring, customer service, fraud detection, and risk management across sectors.

The financial sector is data-intensive and among the most exposed to artificial intelligence. The application of AI in finance is significantly changing how markets operate, risks are managed, and consumers interact with financial services. 

The use of AI in finance is not something new. Traditional analytics have been applied in various functions throughout the financial system. 

For example, AI models have been used for rule-based risk analysis in financial intermediation, risk management and portfolio optimization in asset management, and fraud detection in payment systems. 

In particular, the emerging generative AI technology can generate and execute transactions, even without human intervention. 

It enables the processing of huge amounts of data at a speed far beyond human capacity. Generative AI thus offers vast opportunities for the financial sector across several functions, including financial intermediation, insurance, asset management, and payment systems. 

Financial institutions have also used generative AI to strengthen credit scoring, back-end processing, customer support, risk analysis, robo-advising, and know-your-customer processes. 

These four areas offer interesting opportunities for AI in finance:

Financial intermediation: Traditional analytics focus on rule-based risk analysis and fostering greater competition. With the adoption of machine learning, financial institutions have improved credit risk analysis, reduced underwriting costs, and expanded financial inclusion. Generative AI takes this further by enabling enhanced credit scoring using unstructured data, streamlining back-end processing, and improving customer support.

Insurance: Traditional analytics support risk analysis and market competition. Machine learning introduces better risk assessment, lowers processing costs, and enhances fraud detection capabilities. Generative AI enhances risk analysis through the ability to process newly legible data and facilitates easier compliance with regulatory requirements.

Asset management: Traditional analytics help with risk management, portfolio optimization, and high-frequency trading. Machine learning allows the analysis of new data sources and continues to support high-frequency trading. Generative AI contributes through robo-advising, asset embedding, the development of new financial products, and improved customer service.

Payments: Traditional analytics are primarily used for fraud detection. Machine learning introduces new liquidity management tools and strengthens fraud detection. Generative AI enhances know-your-customer and anti-money laundering processes, increasing the efficiency and accuracy of identity verification and transaction monitoring.

To maximize the net benefits for finance, AI regulations must strike a balance between innovation and safety.

While AI has created numerous benefits for the financial sector, there are some challenges related to its adoption. In particular, there are new risks associated with the use of generative AI. 

Since AI can be adopted across different functions, processes, and applications, financial systems will likely become more vulnerable to cybersecurity threats. 

Further, generative AI models are prone to the garbage-in-garbage-out problem, as they tend to capture and sustain the biases and errors inherent in the underlying data that they have been trained on. 

AI models could also generate hallucinations, which are false or misleading information resulting from incorrect or insufficient training data and faulty assumptions. 

The use of generative AI can also create systemic risks. The domination of AI supply chain by a few big tech players results in more uniform behavior. This means that failures and disruptions within the AI systems of big tech players can have widespread effects that lead to overall financial instability.   

Therefore, the key challenge is to build AI regulations that recognize both the risks and benefits of AI adoption. This would help maximize the benefits of AI for finance while minimizing its risks. 

The principles underlying AI regulations must encompass social and environmental well-being, transparency and accountability, and fairness and protection of privacy. 

Given differences in countries’ level of development and extent of AI adoption, global cooperation on AI regulation is also important.

The adoption of AI can deliver potentially large benefits for the financial sector. However, AI also poses systemic risks and potential market disruptions. 

To maximize the net benefits for finance, AI regulations must strike a balance between innovation and safety. Doing so requires international cooperation, transparency, and adaptable principles that can keep up with fast-evolving AI technologies.

Source: blogs.adb.org