The era of generative AI: Driving transformation in banking

It’s been an incredible five months since OpenAI’s release of ChatGPT in November 2022. In collaboration with OpenAI, Microsoft is leading the generative AI wave, first with the announcement of Azure OpenAI Service and Bing Chat, and more recently with the announcements of AI-powered copilots for use across our entire platform of solutions, including Microsoft 365, Dynamics 365, GitHub, and Security.

In the financial services industry, there has been greatly heightened interest in AI and the transformative opportunities of this new wave of breakthroughs. With the industry’s focus on managing risk and generating returns, generative AI has the potential to drive marked improvements in employee productivity, operational efficiency, and customer experience. In my current conversations with customers and partners, I’m asked how Microsoft can help businesses and organizations get started on the journey. This is especially true in banking, where technology is playing an increasingly decisive role in addressing a broad range of financial, regulatory, and competitive challenges. Generative AI has added a whole new dimension to what we mean by intelligent banking and the possibilities it creates to unlock greater innovation and business value at an accelerated pace.

Reimagine banking

Learn how the financial services industry is innovating and transforming.

Discover more

What generative AI means for banking

As generative AI capabilities become available to everyone, banks and other institutions will want to build intelligent solutions to provide revolutionary new capabilitiesfirst with their employees and, over time, for their customers. Microsoft will help enable this with Microsoft 365 Copilot, which we announced last month. Integrated across Microsoft 365, this copilot will provide generative AI capabilities to the entire Microsoft productivity suite.

For banks that want to rapidly apply intelligence to improve operations and drive efficiencies, they can deploy Azure OpenAI Service to accelerate the deployment of their use cases. Azure OpenAI Service provides the OpenAI foundational large language models (LLMs), some of which can be fine-tuned to specific needs for a variety of use cases. Critically for banks, it is deployed on their Azure tenants so that all dataincluding training data and contentstays within the bounds of their organizations. By being fully integrated into Azure, banks also get all the advantages of enterprise-grade security and role-based access included. You also get the benefit of building on the Microsoft Cloud platform where we are infusing AI into all our products, making it easier to integrate these new capabilities into your applications. And it’s only getting better, as we recently announced the availability of GPT-4 in a preview release. GPT-4 is OpenAI’s most advanced LLM, enabling you to drive insights with greater accuracy than previous LLMs developed by OpenAI.

Azure OpenAI Service lets you deploy large, pre-trained, foundational models developed by OpenAI while also enabling you to train them on your data. This means you can potentially transform important tasks such as:

• Writing assistance and content generation.
• Reasoning over structured and unstructured data.
• Summarization of reports and text.

Use cases in banking

As we engage with our customers, we are seeing powerful new use cases emerge. For any scenario, we believe human agency and supervision are critical to ensure that generative AI is empowering and enabling human creativity. In our own products, we enable this through our copilots. The copilot is there to support you and work under your direction, with the human in charge. For example, GitHub Copilot provides code suggestions for developers as they enter code right inside the developer environment. In other situations, a chat-based experience will be a better fit when you are looking to embed knowledge search capabilities or a search bar to accept a prompt to generate some new content. Key banking use cases where generative AI can have the greatest impact include:

• Empowering contact center agents.Generative AI enables you to summarize conversations and get insights across various conversations. Customer sentiment can be measured from the start to the end of the conversation. In addition to summarization, generative AI can provide coaching to contact center staff in real-time, and partially automate the customer journey with human supervision of the next step. It can also feed new intelligence into the contact center knowledge base to enable agents to respond faster to future questions. For all these capabilities, we can aggregate insights for tracking key performance indicators (KPIs) for customer satisfaction, engagement, and impact to the Net Promoter Score (NPS), all of which can be used to continuously improve the experience for customers.
• Empowering advisors.Financial products have extensive documentation that can be difficult to search, making it a challenge to get to an answer quickly. In some scenarios, advisors are certified in their product knowledge. Many banks are exploring the opportunity for generative AI to help advisors retrieve the answers they need from financial product documentation. Generative AI makes it easier to do this through powerful summarization and contextualization capabilities. It can even summarize the key attributes of products in a comparison table. In addition to advisory roles, these enhanced knowledge search capabilities can be built once and used by multiple roles across the bank such as branch staff and contact center agents.
• Content generation. Banks are exploring how generative AI can accelerate the development of content such as pitch books. Pitch books are used by investment banks to generate a proposal for a capital raise or merger and acquisition for an institutional investor. Pitch books are developed collaboratively with content from multiple sources such as an overview of the client, the deal strategy, and marketing materials. For every content generation scenario, human oversight is critical to ensure the quality and accuracy of generated content.
• Code generation.GitHub Copilot was released last year, and developers can now take advantage of generative AI to provide code suggestions for dozens of programming languages, access application programming interfaces (APIs) faster, and accelerate software development. In March 2023, we announced the upcoming GitHub Copilot X, which is trained on GPT-4 and brings AI capabilities to the entire development lifecycle.