The “data privacy paradox” presents a major challenge for the banking industry, which must balance the demand for personalized experiences with the need to protect customer data. A Capco study in 2021 found 72% of consumers view personalized banking experiences as ‘highly important’. Personalization requires data which most banks often have in abundance and is crucial to tailoring products and services to individual customers. However, data privacy is a growing concern. Razorfish research indicated that 76% of consumers understand data privacy risks, yet 56% still share personal information for personalized experiences. Furthermore, 75% of consumers value personalized experiences, but 68% want more control over their data. Transparency and honesty about data usage can go a long way to help build trust. Some banks are leveraging the data privacy paradox as a competitive advantage by demonstrating how they protect customer data.
Many financial institutions are exploring auto-decisioning technology to expedite loan approval processes. However, the research shows only 42% of financial institutions use auto-decisioning in consumer lending, and those that do only approve a third of their loans with it. In today’s digitally-focused environment, customer expectations for quick loan application processing are not being met by most banks and credit unions, according to research by Cornerstone. They identify four critical stages to increase consumer loan growth: awareness, application, approval, and closure.
Generative AI, with its multiple applications including knowledge mining, summarization, compliance, and chatbots, is increasingly being used in many industries. Companies like OpenAI, Google, Microsoft, Amazon, and Alibaba have developed their own generative AI models. However, the adoption of these technologies by financial institutions is slow due to their risk aversion and the newness of the technology. Uses in the financial sector so far have been limited to areas like AI-powered chatbots and back-office tasks. As the technology matures, wider applications in the financial industry are expected, potentially transforming our understanding of AI capabilities.
SouthState Bank quickly implemented an AI chatbot powered by OpenAI’s ChatGPT to enhance productivity. By digesting approved documents, it provides employees with instant answers to their questions, reducing information search time from an average of seven minutes to less than 32 seconds. The chatbot implementation created significant cost savings and boosted efficiency. However, the implementation also resulted in seven key lessons:
- The fear of the unknown was a significant hurdle during implementation, but careful risk analysis and mitigation addressed most concerns.
- The development team ensured the bot could provide its “thought process,” supporting content, and citations for each answer, reducing the “black box” risk.
- Companies need to optimize their internal document production for generative AI uses to improve the model’s accuracy.
- Dialog management can help constrain the AI output and ensure accuracy.
- Employees must be trained to write quality prompts and understand the model’s limitations.
- AI’s ability to transform the output into an easy-to-use format can be a significant time-saver.
- Risk model management and governance around generative AI need to be strengthened, with an understanding of the strengths and weaknesses of each model.
While currently focused on internal use, SouthState Bank is considering expanding the use of the technology to handle data streams, audio, video, and graphics, with potential customer-facing applications in the future.
AI technology such as ChatGPT has emerged as a potential tool in customer service, offering rapid data processing and predictive abilities that can improve efficiency. There are limitations, however, and this technology cannot yet fully replace human judgment. AI’s effectiveness lies in its ability to catalog and analyze vast amounts of data, but it often fails to fully comprehend context or intention. Trust is a major issue for customer interaction, with AI failing to foster the same level of trust as humans. As AI matures, the view is that it should work in conjunction with human agents, handling transactional tasks while humans handle more complex requests, blending technology and humanity for optimal customer service.
