Banking on AI

In 15 years at Microsoft, Daragh Morrissey has witnessed many changes in the banking industry. None, however, has been quite as transformational as what Generative AI (GenAI) promises. We asked Daragh how AI is shaping the industry.

Can you share some of the most interesting applications of AI in financial services?

In contrast to other industries, AI has long lacked a standout application. However, the emergence of GenAI is flipping the script by making its capabilities more accessible than ever. Through engagement with banks, it’s evident they’re increasingly clear on how they want to leverage AI, given its integration into their daily routines.

Many banks harness AI to tackle repetitive, low-risk tasks that have traditionally bogged down human resources. AI excels in uncovering patterns, correlations, and anomalies within data, paving the way for compelling applications in fraud detection and prevention. Intriguing case studies abound, showcasing the prowess of predictive AI. Take, for instance, one of our partners has developed a solution for self-driving financial management. This innovative tool delves into customer spending behaviors, autonomously guiding them in managing their finances. Its utility became apparent during the pandemic, enabling banks to identify financially strained customers and extend personalized assistance.

Can you describe how you see the role of humans evolving as AI starts to perform certain tasks more effectively?

From a regulatory perspective, it’s paramount to integrate human oversight into any AI deployment. This entails placing the human at the helm while providing them with an AI assistant, or co-pilot, as we like it.

The potential for transformation in this realm is immense. Take the contact center, for instance. AI can analyze an agent’s interactions with customers in real time, offering suggestions based on the customer’s unique circumstances and complete relationship history. This prospect is thrilling because it empowers agents to increase productivity and focus on more meaningful tasks. Consequently, we can enhance customer and agent satisfaction levels, all the while bolstering employee retention rates.

What are the keys to getting a human-led and AI-enabled approach right to accelerate digital transformation?

We’ve gathered invaluable insights from developers, among the earliest beneficiaries of AI technology. A few years back, we introduced a co-pilot solution tailored for developers. Initially conceived as an autocomplete tool for coding, it has since undergone significant evolution. Now, it’s equipped to sift through old code, assist with inline documentation, and proficiently convert variables across different programming languages. By closely tracking its adoption and observing developer interactions, we’ve refined our approach and developed new AI tools that are accessible to everyone right off the bat, such as those integrated into Microsoft 365.

Furthermore, we offer two distinct types of customized solutions. First, low-code options are available through AI Copilot Studio. For use cases that need more custom code, Azure AI Studio empowers developers to craft tailored generative AI applications. Additionally, we provide accelerators that streamline integrating structured and unstructured data with a broad range of foundational models. Developers can choose from the OpenAI models, and open source models from Hugging Face and Mistral. 

How can businesses ensure they facilitate responsible AI?

Financial services firms operate under intense scrutiny regarding the security and privacy of their data, necessitating meticulous handling. We advocate for four pivotal principles that underpin responsible AI: design, governance, monitoring, and training.

Take, for instance, the imperative of eliminating biases within data to prevent instances like denial of service. A noteworthy case arose where a woman had her credit card application declined, despite having a higher score and earnings compared to her husband.

Another illustrative case is our collaboration with Mercedes. By harnessing the capabilities of Azure OpenAI Service and OpenAI foundational models, we’ve empowered Mercedes drivers to use natural language to interact with their car and control car settings like air conditioning. The conversations are protected by our responsible AI filtering to put guardrails around the conversation. 

What advice would you give businesses wanting to be ready for the future?

Historically, financial services firms have grappled with the complexities of gearing up for AI, especially regarding data organization and acquiring the necessary skills.

Fortunately, with the advent of tools developed by companies like Microsoft, the landscape has significantly shifted. Now, deploying AI-enabled solutions and crafting them in-house has become far more accessible. The realm of GenAI solutions is particularly ripe for exploration—I’ve seen technical staff with basic technology skills developing their proofs-of-concept, a scenario that would have seemed improbable in the past.

It’s crucial to recognize that AI is not some distant future; it’s already woven into the fabric of our present reality. And it’s here to stay. Therefore, the sooner firms bolster their maturity in this domain, the better positioned they’ll be. Thankfully, this journey toward maturity can be undertaken securely by leveraging the products we’ve diligently refined and perfected.

And what do you think is around the corner? Where might we see AI having the biggest impact in five years?

I anticipate a growing trend of banks embracing GenAI tools to aid their staff across various tasks. It’s a secure pathway for institutions to acquaint themselves with AI’s potential. As staff become increasingly at ease with AI’s capabilities, I foresee banks advancing to incorporate AI directly into customer-facing interactions. In just a few years, engaging in a natural conversation with our bank will become commonplace. AI adeptly handles most requests—a significant departure from many of the rudimentary chatbots prevalent today. 

But that’s not all. Future iterations of GenAI solutions will unveil capabilities we haven’t even begun to imagine. Reflecting on our strides in the past year alone is remarkable. Without a doubt, an exhilarating future lies ahead.