|
AI Setbacks
Adoption is high. Disruption is low. That’s one of the conclusions of a headline-grabbing report about businesses’ use of artificial intelligence from the Massachusetts Institute of Technology last month.
Despite $30 billion to $40 billion in enterprise investment in generative AI, 95% of organizations are getting zero return, according to “The GenAI Divide: State of AI in Business 2025.” (The researchers interviewed 52 leaders from various industries and reviewed survey data from 153 leaders.)
This isn’t a surprise, at least for anyone in the banking industry. I’ve been doing interviews about the latest iteration of artificial intelligence, GenAI, and its impact on banking. What I’ve learned is that bank directors and leadership teams are taking a cautious approach, which I detail in the upcoming issue of Bank Director magazine, due out on Oct. 8.
There are a lot of governance considerations and risks to mitigate, and thankfully, banks tend to do well in these areas. Just like any other industry, banks are interested in improving productivity through artificial intelligence, as well as growing and expanding capabilities. But there haven’t been a lot of reports about showstopping uses among the nation’s smaller banks, many of whom don’t have the staff to scale up uses internally. Getting help with emails, transcripts and PowerPoint presentations hasn't translated into an earnings impact yet. Many banks are only experimenting with limited use cases that don’t involve customers or customer data.
Smaller banks may see better success rates when they partner with vendors than when they try to implement AI on their own. AI deployments that make a difference also will need solid leaders, not just junior-level staffers experimenting at their desks. Vendors may have proven platforms where smaller banks don’t need to play guinea pig, but here too, there have been problems. “Most [enterprise-grade tools] fail due to brittle workflows, lack of contextual learning, and misalignment with day-to-day operations,” the MIT researchers wrote.
Perhaps a partnership approach would work better than a vendor-controlled approach. Organizations with external partnerships for AI see twice the success rates as organizations that focus on internal builds, the MIT researchers wrote. “Buyers who succeed demand process-specific customization and evaluate tools based on business outcomes rather than software benchmarks,” according to the report. “They expect systems that integrate with existing processes and improve over time.”
• Naomi Snyder, editor-in-chief for Bank Director
| 
