Making the Most of What We Didn’t Sign Up for

Artificial intelligence is one of those things that can either be high stakes or low stakes. On the low stakes end, you might ask ChatGPT to create a suggested itinerary for your European vacation. Or, if you’re a writer for Buzzfeed, you might ask an AI image generator to design 50 different Barbie dream houses to represent all 50 states.

On the high stakes end, you might find a deep-fake video of a world leader making commentary that the real-life version of that leader never made circulating online, potentially influencing the outcome of an election or causing political unrest. You may start hearing about longtime screenwriters’ careers ending because AI has begun writing TV and movie scripts. And, you’ll observe your social media feeds continue to be influenced by AI algorithms that like to show us content that isn’t always truthful or great for our mental health, but is likely to get a rise out of us, contributing to the spread of misinformation and polarization in society.

The potential high-stakes uses of AI are concerning, and have left some of us feeling weary about the technology’s increasingly widespread adoption. But AI is here, it’s growing, and it’s going to be woven into our day to day lives whether we like it or not.

And, despite them not signing up for it, organizations, including credit unions, are now in a position that requires them to interact with AI in order to successfully compete for business. During a recent interview with TV host and comedian Bill Maher, Tristan Harris, executive director and co-founder of the Center for Humane Technology, said, “Whoever jumps on the train gets the benefits of AI, and it starts to confer power to those who start adopting it versus those who don’t.”

So what are the implications of AI on credit unions at this point in the technology’s evolutionary journey? For one, there are numerous potential, beneficial ways to use AI in credit union business. These include building chatbot agents, assisting with lending decisions, writing job descriptions, creating visual and written marketing materials, detecting fraud, making member offers more personalized, and improving overall efficiency and productivity, just to name a few.

The use of AI also poses risks to credit unions in a number of areas, including data privacy, ethics and security. In a Filene Research Institute research brief released in September, “Top 5 Things That Credit Unions Should Consider When Building an AI Policy,” author John Best discussed how to benefit from AI while mitigating the risks, emphasizing first that credit unions must establish a code of conduct or ethical guidelines pertaining to its use. “The potential for AI to deceive or manipulate people is arguably unprecedented and necessitates thoughtful consideration about its deployment strategies,” Best wrote.

Second, Best said, “Credit unions must prioritize the protection of their members’ data by implementing strict protocols for data access, usage and retention.” In other words, when using AI tools, make sure confidential information stays confidential and isn’t let loose via ChatGPT’s inquiry box, for example.

Third, credit unions should watch out for potential bias while training their own AI models. They don’t want, for example, an AI learning model to pick up on common biases in lending and produce unfair decisions during the underwriting process.

Fourth, Best noted that as “vendors rush to integrate AI technology into their products, the development of AI within your credit union may face setbacks if each platform is treated as a separate entity.” Therefore, it’s important to “proactively implement AI governance frameworks.” So, determine who exactly is in charge of AI as it pertains to third-party technology products to “maximize the benefits of AI while maintaining data integrity and achieving organizational objectives.”

Finally and most critically, according to Best, a credit union’s AI policy must thoroughly address security. AI security is more complex than the cybersecurity practices most credit unions are used to – not only does AI system access need to be considered, so does the behavior of the AI platform itself. The research brief described the following as an example: “… imagine a scenario in 2025 where an AI operator has shown significant progress in handling overnight operations, alleviating concerns about unsupervised employees and possible fraud. However, a new challenge emerges when a vendor attempts to exploit the AI system by tricking it into providing unauthorized access to files.”

What’s more, while credit unions may use AI to help them detect fraud, the fraudsters are using AI to their advantage as well by crafting phishing campaigns and other schemes that are more convincing than ever.

It’s worth noting that these five risk areas discussed by the author are not intended to scare credit unions away from AI adoption; in fact, failing to adopt it might be the biggest risk of all. “Ignoring or downplaying the importance of nascent technologies can result in missed opportunities,” Best wrote. “In contrast, understanding their potential, and strategically incorporating them into our plans, will ensure that we are not left behind in this rapidly evolving digital landscape.”

The growth of AI may be controversial, but most can agree that its low-stakes uses can be helpful (or at least good for a laugh) and that it’s most high-stakes uses could be catastrophic. For credit unions, getting familiar with AI tools and devising a strategy for their implementation is essential. However, it’s also important to remember that AI should be used to supplement real life, human-to-human member service interactions – not replace them.

A Personal Note

Speaking of big changes in the digital world and its implications on lives and careers, November 2023 is CU Times’ final print issue and thus this is my final editor’s column to run in a print edition of CU Times. We’re going fully digital heading in 2024, and our plans include the distribution of a new product that launched last month – our Executive Briefing, a monthly compilation of exclusive editorial content and analysis of credit union data, trends and news, all packaged together for subscribers in a monthly email.

This shift is a bittersweet one. On one hand, it takes away a project that I’ve been nurturing ever since I was hired eight years ago as managing editor of CU Times, a role that primarily required me to run the print issue show. I spent countless hours piecing together content for each print issue’s “map” like a jigsaw puzzle; brainstorming stories and topics to fit with each issue’s theme; reviewing and reformatting articles written by many talented freelancers, reporters and contributors; and copy editing numerous magazine proofs. As time went on and we began publishing fewer and fewer print issues, this project – like a growing child – required less and less of my attention.

On the other hand, it represents a big step forward and sets CU Times up to successfully compete with other credit union news organizations in the future. As with AI, we didn’t actively choose this shift to digital – it’s something that evolved on its own, and in order to stay relevant we are required to get on board with it. The fact is, consumers of news aren’t sitting around the breakfast table leafing through the newspaper anymore – they’re taking in content on their computers and mobile devices.

I hope you’ll join us on CUTimes.com, in your email inbox and on our social media pages from here on out, and that you’ll continue to look to us as the top source of credit union industry news and analysis. Thank you for reading over the years!

Natasha Chilingerian is executive editor for CU Times. She can be reached at nchilingerian@cutimes.com.