Financial institutions are investing heavily in AI, with North American banks leading the way in adopting gen AI for enhanced customer service and operational efficiency.

Today, AI's impact is evident across an array of financial applications, from automated knowledge management to personalized banking services. Specifically, North American banks are heavily investing in generative AI (genAI) to drive innovation and improve operational transparency. GenAI is helping enhance banking and chatbot capabilities, offering experiences that are distinct from traditional automated services. GenAI agents are now available 24/7 in multiple languages, to assist customers with banking queries and account management.

To support these efforts, they are acquiring crucial hardware like AI chips and data processing infrastructure, while simultaneously channeling resources to hire human talent.

Finance and AI: A Synergy in the Making

Financial institutions are among the most data-rich organizations, handling billions of customer transactions and interactions across both digital and physical platforms. This wealth of data is essential to train A.I. systems, which is why banks have been pouring significant resources into maintaining the technology.

The goal is to refine existing customer processes and explore high-impact use cases for inculcating A.I., scaling innovative prototypes into robust solutions. For instance, Paris-based Amundi, managing €2 trillion ($2.1 trillion) in assets, utilizes genAI tools to customize investment portfolios for its more than 100 million clients. By engaging clients in discussions about their risk preferences through its chatbot, the asset management firm crafts portfolios that meet individual needs while also gaining real-time insights into client sentiment.

“Younger demographics, who will make up an increasingly large percentage of overall target customers for the finance industry - are very comfortable interacting with chatbots and often prefer this option over traditional options like call centers,” Ranjit Tinaikar, CEO of Ness Digital Engineering, told me. “AI can drastically improve the risk insights in pricing consumer loans. The banking industry is tapping into these trends to future-proof the way they deliver customer service and optimize experiences”

Renowned North American investment bank and wealth management firm Morgan Stanley, recently introduced its new internal genAI model designed for its financial advisors and support staff. Powered by OpenAI’s GPT-4, the ‘AI @ Morgan Stanley Assistant’ helps thousands of wealth managers to swiftly retrieve information from an extensive internal knowledge base. It can also summarize client meeting notes and automatically draft follow-up emails.

Despite its promise, the use of AI for handling sensitive financial decisions does have its drawbacks. AI models’ decision-making abilities are often hindered by unreliable data or unexpected infrastructural errors, which in turn impacts their effectiveness.

Challenges and Limitations of AI in Financial Services

A.I.'s ability to disrupt industry dynamics appears closely tied to the nature of customer data and the challenges it addresses. In banking service areas like wealth management or lending, where data is less abundant and unstable, A.I.'s impact has been more incremental than transformational. The finance industry’s experience with AI also raises questions about whether AI will democratize or consolidate power within the sector.

“Most banking corporations don’t have the requisite organizational expertise and governance to deploy A.I. solutions. GenA.I. is only as smart as the data used to train the large language models (LLMs),” Tinaikar added. “Access to this data, which is often fragmented across internal and external systems, is a major challenge to create reliable financial A.I. models.”

It is crucial to ensure that A.I. operates within a secure, ethical, and compliant framework to prevent potentially sensitive data misuse. In this context, larger companies with vast resources for technology and data investments hold a notable edge over their smaller rivals. For small and medium-sized businesses (SMBs), the real challenge to creating smarter and reliable A.I. systems is their inability to afford A.I. experts and invest in security management architectures.

“You do need a “human-in-loop” for sensitive decisions around A.I.-generated pricing and risk recommendations,” said Tinaikar. “In the interim, SMB corporates have to develop their self-governance to protect their consumers, investors, and other key stakeholders.”

Why Human Judgment Remains Crucial in AI-Enhanced Finance

GenA.I. represents a transformative shift, enabling banks to fully harness the power of their customer data. By analyzing structured and unstructured data from diverse sources and understanding the relationships between them, it offers enhanced visibility and insights into both risks and opportunities within and beyond the bank. But, as AI makes data analysis more efficient and cost-effective, the role of human judgment becomes even more crucial.

AI systems currently lack the nuance and creativity of human insight, making them unreliable for sensitive use cases. Although AI is an incredibly powerful tool, financial managers and institutions stand to gain the most from combining AI's power with unique human abilities.

“The difference lies, of course, in the type of emotional judgment applied by human experts,” explained Tinaikar. “As AI introduces the potential for a disruptive shift where machines could replace human judgment, we need to develop regulatory and technological solutions to ensure that humans remain involved in mission-critical tasks where machine-simulated judgment may not be reliable.”