Exploring the impact of generative AI in financial services

More than $136 billion. That was the global artificial intelligence (AI) market’s size in 2022 — and, as noted by Forbes, expected growth for this sector will exceed 37% year-over-year through 2030 as AI tools evolve and enterprise investment ramps up. For the financial services sector, AI offers multiple benefits, including improved trade monitoring, enhanced credit analysis and reduced costs. According to Fortune, AI has already helped 36% of financial firms reduce their total costs by 10% or more. More recent advances in AI include generative tools capable of producing human-like responses to complicated, multifaceted questions. Used effectively, these solutions can help financial organizations chart a new operational course. Here's how.

Exploring the impact of AI in financial services      Download the e-book

The basics of generative AI for finance

As a concept, generative AI (GenAI) is simple: ask a question and get a relevant answer. We're familiar with this function as human beings — using the sum of our experience paired with relevant, contextual data, we can (mostly) provide accurate answers to questions.

However, for GenAI tools, context doesn't come standard. While asking questions leads to human-like answers, the accuracy and relevancy of these answers depend on the data available, the questions asked and the parameters set by users.

Consider GenAI juggernaut ChatGPT, which passed the 100-million-user mark just two months after its release. It leverages both user queries and the data accessed by those queries to refine its AI model. In effect, it's constantly learning, adapting and refining its contextual model. And its database is expanding — as of September 2023, the "browsing" feature of ChatGPT lets it access data from the Internet at large.

Choosing the right tools for generative AI

“When all you have is a hammer, everything looks like a nail.” It's an old saying but a good one: The tools you have available influence your perception of their use.

For banks and GenAI, this speaks to the need to find the right tool for the job. Consider a financial services business looking to improve its customer targeting. In that scenario, using a public solution such as ChatGPT to access public data and create an ideal client profile makes sense. The sheer volume of information available online can help companies pinpoint general patterns that can inform new business strategies.

However, if businesses want to combine disparate financial data resources to inform ongoing strategy, using a public tool could result in accidental data exposure. Here, solutions such as ChatGPT Enterprise or Azure OpenAI can help.

In the case of Azure OpenAI, businesses can access a version of ChatGPT 4 that's entirely run in their own tenancy. As a result, they're able to access the intelligence of GenAI without exposing critical data to prying eyes.

Generating insight: Use cases for GenAI in financial services

With the right application of generative AI, there are several beneficial use cases for financial services, including:

Improving customer service

Using a combination of demographic, social media and user-provided customer data, GenAI tools can help businesses improve customer service. For example, GenAI tools can suggest email improvements that help make messaging more accurate or engaging and also schedule follow-up tasks to help staff stay in touch with clients.

Enhancing data analysis

The volume and variety of financial data is increasing. Firms now can make use of on-site sources, mobile and web app data, and publicly available information in the cloud.

Effectively analyzing this data, however, is challenging for even the most experienced teams. By combining prompt-based interfaces with continual learning frameworks, GenAI tools can help financial firms pinpoint relevant data and discover key trends. Post-analysis human oversight, meanwhile, can refine prompts to help enhance data analysis over time.

Improving fraud detection

Patterns of behavior can indicate fraud. However, the scope and scale of these patterns can be problematic for service organizations to track. Secure GenAI frameworks make it possible to consider the entirety of user actions rather than a simple subset — which in turn improves the ability to detect fraud more quickly.

The role of human insight in GenAI

GenAI represents a fundamental shift in the way financial firms do business, but even the most advanced artificial intelligence tools aren't stand-alone solutions. To make best use of generative tools, there's a corporate constant: the need for humans in the loop.

Consider Microsoft Copilot, which the company describes as "your everyday AI companion." Released Nov. 1, 2023, the newest version of the solution includes support for Microsoft 365.

Or look at OpenAI's new advanced analytics tooling that allows companies to paste financial data into prompts to deliver analysis, provide trend mapping or even produce Python code. Another option is ChatGPT Enterprise, which prevents the public sharing or use of confidential financial data.

The common denominator in both cases? Human beings. No matter the GenAI tool chosen, generated outputs depend on data inputs, which must be selected by knowledgeable users.

A human-in-the-loop review of AI-generated content serves as a safeguard against inaccuracies, irrelevance, and unintended biases.

Meanwhile, when it comes to the multifaceted usage of Copilot, it's the actions of users that inform generative AI responses. For example, if staff are creating client correspondence documents in Microsoft 365, AI can provide predictive text, proactive suggestions and proofreading analysis to ensure critical details aren't missed.

In other words, it's telling that Microsoft called it Copilot and not Autopilot. While GenAI tools have been evolving at breakneck speed, advancement doesn't always equal accuracy. Instead, useful output depends on the questions asked, parameters given and data used. As a result, solutions like Copilot always offer the chance for user review before emails are sent or documents filed.

Improving intelligent security with the SAFE AI Framework™

No matter which GenAI tools financial services companies choose, data safety remains the top priority. Even accidental data exposure can prove problematic for companies as compliance expectations evolve. Whatever third-party programs promise or deliver, finance organizations are ultimately responsible for compliant data use.

Mazars' SAFE AI Framework can help financial firms navigate the new, generative landscape. Focusing on secure infrastructure, adaptable processes, factual data sources and ethical governance guardrails, the SAFE AI Framework sets the stage for GenAI implementation that avoids common pitfalls and lives up to its potential.

Ready to make the most of generative AI? Start with Mazars.

Download the e-book now

The information provided here is for general guidance only, and does not constitute the provision of tax advice, accounting services, investment advice, legal advice, or professional consulting of any kind. The information provided herein should not be used as a substitute for consultation with professional tax, accounting, legal or other competent advisers.