Friction is actually good for GenAI projects: How credit unions can overcome it

A recent MIT study revealed that nearly 95% of GenAI projects fail to meet their goals. That number sounds alarming but it’s not a reason for credit unions to step back. It’s a signal to step smarter.

MIT researchers found that the biggest reason for these failures wasn’t poor technology—it was friction: the invisible resistance that slows, misdirects, or derails AI initiatives. Ironically, many organizations that failed had one thing in common—they avoided friction altogether, trying to make their AI journey “smooth.” But in innovation, friction isn’t your enemy; it’s a mirror. It exposes what needs fixing.

For credit unions, which operate on trust, empathy, and compliance, friction is not only inevitable but it’s valuable. By identifying and working through it, you build stronger alignment between technology, mission, and members. With extensive experience helping financial institutions use data and AI responsibly, we’ve identified five key types of GenAI friction along with practical ways to overcome each.

MIT’s research shows that many AI efforts fail because they’re launched as experiments, not business solutions. Credit unions fall into this trap when they adopt GenAI “to keep up” rather than to solve a defined member or operational challenge.

To overcome this, every AI initiative should begin with why. Tie it directly to a measurable outcome whether it’s reducing call center volume, improving loan turnaround time, or deepening member engagement. Define success metrics early, link them to your mission, and ensure leadership, operations, and compliance teams are aligned.

MIT highlights another failure driver: the gap between AI’s potential and an organization’s data readiness. Credit unions often hold rich data—core transactions, CRM records, collections history but it’s scattered across systems. Without clean, connected data, even the best models will misfire.

The key is integration and context. Clean your data, establish strong governance, and involve your domain experts early—your collections officers, loan processors, and member service teams. They know what “good” looks like. When AI learns from real-world workflows, it produces insights that are accurate, trusted, and actionable.

Even when GenAI models work well, many projects fail to scale because they never fully fit into daily operations. MIT’s analysis points to this “last-mile friction” as one of the biggest killers of AI ROI.

For credit unions, the fix is practical: design AI around how your teams actually work. Map existing workflows and identify points where AI can enhance not interrupt them. Whether it’s a GenAI chatbot assisting members or a collections recommendation tool for staff, integration with your core and CRM systems is critical. When AI fits naturally into the flow of work, adoption soars.

MIT also found that organizations underestimate the human side of AI. For credit unions, where culture and empathy are at the core of the brand, this friction can make or break success. If staff view AI as a threat or don’t understand how to use it, they’ll resist it—no matter how powerful it is.

The solution? Transparency and empowerment. Communicate the “why” early, train your teams to interpret and use AI output confidently, and design solutions that augment human judgment, not replace it. When employees see AI helping them serve members better, friction transforms into engagement.

Finally, the MIT study underscores that many AI projects collapse at the governance stage not because the models failed, but because organizations couldn’t move them safely into production.

For credit unions, this friction often stems from compliance, fairness, and auditability concerns. The answer isn’t to slow down innovation, it’s to embed responsible AI principles from the start. Define ownership for model oversight, involve risk and compliance teams early, and document every step. A well-governed AI strategy doesn’t stifle progress, it makes it sustainable and scalable.

Every point of friction tells a story about where systems, people, and goals aren’t yet aligned. For credit unions, acknowledging that friction is part of progress builds resilience and trust.

Those who accept it, study it, and work through it become the ones that truly harness AI’s potential, to personalize member interactions, streamline operations, and make smarter, fairer decisions.

At AiVantage, we believe GenAI success for credit unions isn’t about chasing a “frictionless” future—it’s about mastering friction-aware transformation. Because the institutions that lean into the tough spots today will be the ones leading the movement tomorrow. Reach out to us today to start building your GenAI success story.