Open discussion: AI strategies for banks

Artificial intelligence (AI) is transforming the banking industry, providing unparalleled opportunities for innovation, operational efficiency and customer service improvements. However, these opportunities do not come without challenges, such as data security, compliance and ethical considerations. 

These topics were recently discussed in a webinar, AI Strategies for Scalable, Secure and Compliant Banking, hosted by Finextra. During this session, Red Hat vice president and global head of financial services, Richard Harmon, and product manager, Will Caban, joined the CTO of Dwolla, Skylar Nesheim, and Finextra’s Sharon Kimathi to explore best practices and tools for AI integration in banking, highlighting the importance of open source models, generative AI (gen AI), the use of synthetic data and regulatory frameworks.

Experimentation and exploration

In a poll taken during the webinar and published in the post-webinar report, 58% of survey respondents supported the position that most financial services institutions are still in the experimentation or exploration phase. While trends and use cases can vary from organization to organization, clear themes about how banks are experimenting with AI arose during the session: 

• Generative AI: Often viewed as delivering impact to the customer, providing rapid responses to inquiries. Gen AI can also be a powerful tool for market analytics and the development of new products
• Financial crime prevention: AI systems can analyze vast amounts of data to identify patterns and anomalies indicating fraudulent activity. Robust fraud detection protects the bank’s assets and enhances customer trust and confidence
• Payment processing: AI can simplify and optimize transactions more efficiently, leading to more secure, faster and cost-effective transactions, benefiting consumers and institutions
• Agentic AI: AI agents performing tasks autonomously hold promise for revolutionizing banking. AI agents monitor systems, identify issues and take corrective actions, improving operational efficiency and customer experience

Synthetic data and the regulatory conundrum

Synthetic data is crucial in AI–especially in banking, where data privacy is vital. Generated by algorithms, it replicates real data without revealing sensitive information, enabling new data sharing and analysis. This enhances security and fosters innovation, as demonstrated in financial crime detection, where institutions can collaborate on models without risking customer data. 

Synthetic data greatly enhances valuable data sharing among institutions and countries, strengthening their collective efforts. It's crucial for AI training, especially considering the high costs and limited access to real data. With synthetic data, organizations can quickly generate large datasets for training their advanced AI systems, boosting scalability and affordability. Plus, it helps reduce biases found in actual data, leading to more accurate and fair AI models. 

Balancing regulation and innovation is essential for the ethical deployment of AI. New regulations like the EU Artificial Intelligence (AI) Act emphasize data privacy, transparency and robustness, reducing risks and building trust. Effective regulation should protect innovation while ensuring safety, necessitating a thorough understanding of the technologies involved. 

The industry must develop frameworks that support innovation while enforcing safeguards against misuse. In another poll, 41% of webinar attendees cited compliance concerns as the reason behind fully adopting AI-driven services. 

Best practices to consider

While AI best practices evolve rapidly, there are a few guidelines that banks should consider. One of the most important guidelines to remember, according to Caban, is to “stop chasing the next thing.” 

“Because here’s the thing: it doesn’t matter what that next big thing is. Someone will copy it, because nowadays, it’s very easy to do,” said Caban. Instead, focus on customer pain points and internal pain points, and do it in an iterative way that brings sustainable AI innovation to your organization.” Sage advice.

Other timeless best practices include:

• Align AI to banking use cases, data management and governance to build compliant and safe enterprise AI at scale
• Consider using open source AI models, weights, algorithms and frameworks to provide transparency, community support and the ability to help comply with regulations like the EU AI Act
• Advocate for a balanced approach to AI adoption, including focusing on solving specific problems, ensuring regulatory compliance and fostering a culture of empathy and collaboration

Conclusion

AI in banking has unique benefits and challenges. Improved efficiency and customer experience increase interest in new use cases. Banks must align AI adoption with responsible practices to maximize growth and customer satisfaction. Embracing AI tools, using synthetic data and following regulations are key to unlocking AI’s potential. Addressing ethical issues and biases in regulatory environments enhances efficiency and innovation. By focusing on customer needs and fostering empathy, the financial sector can achieve growth and innovation with the right strategies and safeguards.

Red Hat is committed to delivering innovative AI-driven solutions with the financial services industry. Red Hat AI helps accelerate the adoption of production-ready services. Visit our financial services page to learn more. To learn more about how Red Hat can help you generate custom LLMs using Red Hat Enterprise Linux AI and a synthetic data generator (SDG), check out our interactive demo.