Trustworthy Artificial Intelligence. This new directive will likely boost efforts already underway to police the use of AI with an emphasis on preventing privacy and discrimination-related concerns surrounding personal data and financial transactions. In January 2023, the National Institute of Standards and Technology (NIST), the nation’s leading authority on standards and technology specific to the nation’s economic security, issued a new Artificial Intelligence Risk Management Framework and a companion playbook. This is the first comprehensive attempt at industry guidance, and it is an effective tool; however, this framework will likely be updated as organizations evaluate its effectiveness. An American Banking Association Journal report quoted Acting Comptroller of the Currency Michael Hsu as urging banks to be cautious in implementing AI products. Hsu urged bankers to talk with their regulators as they consider AI products “rather than engaging with them afterward.” That is sound guidance, even if, at this point, regulators are at best only in the comment stage with any requirements around AI. We know the marketplace is ripe with competitors who may have the means to introduce AI products, so we assume your bank will, at some point, introduce AI products. But before you do, here’s our second point. Develop a clear AI strategy for your bank. What problems do you think AI can solve at your bank? How can AI help your bank? What exactly would AI help you do faster or more accurately? The Large Language Models that have garnered marketing cache (e.g., ChatGPT, Bard, Perplexity) use deep learning algorithms that can recognize, summarize, translate, predict and generate content using large datasets. But there are dozens, if not hundreds, of niche AI products that solve particular business problems. Consider where you need help with bottlenecks or unsatisfactory customer experiences. What is routine — and time-consuming? Where do you need answers quickly about trends? And decide if an AI tool can fill that gap in your business process. Common suggestions include fraud detection, generating compliance reports, business intelligence reporting (e.g., tracking loan trends) and customer service. But what specific issues does your bank face? Are there other ways to solve these problems? Develop policies for the use of AI at your bank. Are you concerned that your employees already use AI tools to help them with their work? Have you asked them? Have you or your legal team reviewed the terms and conditions of these commercially available tools like ChatGPT? What about the fear of using “poisoned data” or data that has been corrupted or altered in misleading ways? Who owns the data you type into the tools? These are all points that a well-crafted AI policy should address. What precautions will you take to prevent bad actors from using AI against your bank and customers? What controls can you enact to keep a criminal from impersonating a customer and fooling your AI platform? These are all questions you should consider and address in your policies and procedures before adopting any new AI-based technologies at your bank. Consider using outside expertise. Your community bank is likely already facing pressures from larger competitors, who presumably have internal staff who are experts in AI for financial institutions. You can benefit by leaning on reputable fintechs, including your core processor. They have the resources to test and refine AI software to meet banking requirements. And they understand the business problems you are trying to solve. But beware — there are many startup companies in the marketplace offering AI-powered fintech for banks. With recent interest rate changes, many have or are quickly running out of capital. Exercise good judgment during due diligence. Even then, consider consulting with a company, such as ImageQuest, that routinely assesses IT and AI services for our clients. We can help you choose the correct solutions for your bank’s specific issues — and, in some cases, help curate the best tool(s) for your bank. The Trough of Disillusionment Gartner, a leading research and advisory firm, has coined the phrase “Gartner Hype Cycle” to represent the social reaction to new technology. The cycle includes the “Technology Trigger,” introducing a new or breakthrough technology, followed by the “Peak of Inflated Expectations.” AI hit peak excitement in Q1 2023. Now AI is in Gartner’s “Trough of Disillusionment,” where a new technology fails to deliver on its early promise. To be sure, the Hype Cycle continues as more people work with and understand this emerging technology, leading to “Mainstream Adoption.” But for now, we don’t want our community banker friends disillusioned over costs, mistakes and disappointment with their plunge into AI. If you follow the measures above, you will be ahead of the curve. 9 West Virginia Banker
RkJQdWJsaXNoZXIy ODQxMjUw