Executive Summary
This study draws upon a series of in-depth interviews conducted with credit union executives, vendors, and data scientists regarding their experiences with AI and financial services.
This report also introduces design thinking principles that can be used to broaden AI discussions. AI tools always exist in relationship with people, both the members we hope to better serve and credit union staff whose works will be affected by these new tools.
In order to address this human component, applying design principles of (1) desirability for the end users; (2) viability as a business plan; and (3) technical feasibility can transform the assessment and implementation of AI beyond an oversimplified "tools and technology" framework.
What is the research about?
When considering applications of artificial intelligence in financial services, design thinking ensures a process that takes desirability, viability, and feasibility into account.
Two credit unions interviewed for this study wanted to improve member financial well-being with better aligned products and services to meet member needs. Their goal was to leverage AI to better serve member needs from behind the scenes.
During the design process, the credit union learned that it lacked linked data about its members to feed an AI solution. They also lacked a clear, organization-wide data governance strategy to guide how data would be aggregated, where it would be housed and how security could be ensured.
What are the credit union implications?
When using AI to serve member needs, there may be discomfort from members which may be attributed to the "uncanny valley" effect. This term refers to the uneasy feeling experienced when AI looks "too human."
In studying "end user" desirability, remember to include both members and credit union staff as end users.
- How does the prototype affect members? Does it create desirable improvements for them?
- How does the prototype affect staff in terms of work roles, processes, policies, and responsibilities? Does it create desirable improvements for them?
In addition to end user uptake and use, pay attention to obstacles, areas of friction, and unintended effects.
Explore AI through the design lens to expand the conversation beyond technology and prompt consideration of the end user's needs and preferences.
Build a data governance model to best position your credit union for making uses of AI further down the road. AI comes after a data governance model has been built out and implemented.
Learn about your member needs.
- Put yourself into the shoes of the member and identify their priorities
- Understand needs across different member segments and engagement channels
Begin with low-hanging fruit: improve laborious processes first.
- Automating more mundane processes may end up being more impactful: streamlining operational processes from cybersecurity, to compliance and administration, to fraud detection can also improve staff buy-in by providing them with time to focus on other priorities.
Remember your responsibility to use new technologies to protect members by monitoring its use and impacts. At the same time, take this opportunity to become leaders in the transparent, fair, trustworthy and responsible use of data in financial services.