AI in Banking: Cautious Steps Forward
Banks aren't exactly racing with AI; they're inching forward. With AI agents entering the picture, you'd think the buzz would be about groundbreaking deployments. Instead, the story here is more about banks holding their cards close to the vest. Agentic AI, with its ability to plan and act without a human babysitter at every step, is being treated like a potential liability instead of an opportunity.
Limited Power of the Agents
We're talking about AI agents that could be doing a whole lot more than they are right now. A whopping 73% of banks are letting these agents out of the gate but only for a select few tasks, making them just 10% or less of the AI portfolio at the lion's share of institutions. Translation: banks are tiptoeing around their AI investments, probably because they don't have the monitoring muscle to handle these agents.
"As AI scales and becomes more autonomous, the next challenge is clear: turning governance from a framework into an operating capability," notes Anne Kleppe, co-author of the report.
Risky Business: Governance vs. Speed
Some might say better safe than sorry, and it's easy to see why when nearly half the banks have been bitten by AI-related blunders in the last year alone. But the real kicker is the stubborn belief that stepping on the governance gas will slow down the race to innovate.
Yet, the data flips this on its head. Banks with tighter risk management are 2.5 times more likely to be cruising the advanced AI highway. Standardized, pre-approved routines for lower-risk activities are key, enabling them to shift gears from brainstorm to production in under six months. Smart cookies don't sweat the small stuff—they reserve energy for high-stakes challenges.
The Missing Piece: Operational Control
Walking the talk is where many banks stumble. Most have set up the basic scaffolding to keep AI risks in check. Almost everyone (96%) aligns AI risk with existing frameworks, but day-to-day control tools are still catching up. Imagine having a map but no road signs—that's where these banks are stuck.
- Risk Appetite: 73% know what risks they’re willing to take, compared to 57% last year.
- AI Policies: 77% have penciled down a dedicated policy.
- Central Control: Less than half have a centralized, tool-based AI inventory.
The Path to Full-Scale AI Adoption
The road ahead isn't a straightforward path. Full-scale AI adoption means expanding the scope and depth of control mechanisms. This is where the fate of AI in banking teeters between revolutionary potential and managed risk. Banks managing to straddle this line can expect to see AI accelerate their operational capabilities in ways traditional systems never could.
Anne Kleppe from BCG throws down the gauntlet, pointing out that with agents being a small part of portfolios, banks have a golden window of time to bolster their oversight systems before AI agents become deeply entrenched in their operations. It's a clear call to action: nail down governance now, or risk stumbling when AI starts playing a bigger hand.
In the end, the ace up the sleeves of these financial giants will be their ability to convert governance from a mere checkpoint into an enabler of agility and innovation. Those banking on AI must balance speed with precision, turning the promise of technology into tangible value without tripping over their own shoelaces.
