Most banks are pouring money into AI but failing to deliver the outsized returns their boards expect. EY's 2025 AI Confidence Pulse research found that only 40% of banks consider themselves AI leaders, and a separate survey showed 98% have taken financial hits from AI-related risks like hallucinations, data bias, and security gaps. A framework built on five core attributes can move institutions from scattered pilots to coordinated, value-generating programs.
Five attributes that define effective AI strategies
The first is a bold, coordinated vision that breaks down silos. Top performers improve their AI maturity at 2.3 times the rate of peers, according to the Evident AI Banking Index. "Without a platform-based approach, banks risk creating 15 versions of the same AI function - just like they once used 15,000 spreadsheets," said Sameer Gupta, EY Global Financial Services AI Leader.
Strong governance is the second attribute. Fifty-two percent of banks call it their top challenge. The October 2025 EY Responsible AI Pulse Survey showed that firms with formal oversight committees and real-time monitoring are far more likely to see revenue growth and cost savings. Yet nearly every respondent reported financial losses from AI failures.
Third, AI investments must be business-led, not IT-led. "AI isn't a backlog of use cases; it's a capital allocation decision," said Preetham Peddanagari, EY UK&I Chief Technology Officer. He urged banks to fund the few bets they can govern end to end and cut everything else. Forty-four percent of banks struggle to prioritize the right use cases, the EY research found.
The fourth attribute is a human-centered approach. "The biggest barrier to scaling AI isn't the algorithm - it's the change management," Gupta said. Training teams and reworking processes often takes twice the effort of building models. An EY/MIT report found that three-quarters of executives now see agentic AI as a coworker, while 84% of desk-based workers are enthusiastic about the technology. At the same time, 56% worry about job security - a paradox leaders must address. AI Learning Path for CEOs resources can help senior leaders build the skills to guide that shift.
Finally, long-term thinking matters. Banks need modular, scalable infrastructure that can absorb rapid changes in AI capabilities. Hybrid approaches - cloud for most workloads, on-premises large language models for the most sensitive data - are gaining traction as institutions balance risk and return.
From pilots to platform: the governance imperative
Governance gaps don't just create regulatory risk; they destroy value. A governance watchtower - automated controls testing, human oversight, standards for explainability, and continuous model validation - has emerged as a leading practice. Gupta said clear ownership of data security, vendor management, and model design is essential: "Standardization is the way to scale responsibly."
Beatriz Sanz Sáiz, EY Global AI Sector Leader, framed the core challenge: "The challenge isn't a lack of AI use cases, but narrowing the options to those that generate real value, measured against financial returns and strategic goals." She said a platform-based strategy avoids fragmentation while still giving business units room to experiment. Enterprise AI agent libraries that reuse logic for fraud detection, forecasting, and regulatory filings can cut down the sprawl that hurts ROI.
Recommended actions for the C-suite
EY advisors recommend that banking leaders move immediately on several fronts. Set a clear vision that ties AI directly to EBITDA and other bottom-line metrics. Shift accountability for outcomes from IT to business heads. Establish a governance watchtower with participation from technology, risk, legal, and compliance. Prioritize data lineage and quality, including third-party data used in large language models. And invest continuously in upskilling programs and change management to build AI fluency across the workforce. AI for Executives & Strategy resources can support these upskilling efforts.
Why this matters for Executives and Strategy
AI is not a technology project to delegate. The research shows that only a fraction of banks are turning AI investments into revenue and trust. Executives who treat AI as a capital allocation decision - funding a small number of business-led, governed, and measurable initiatives - will widen the gap with competitors still running fragmented pilots. The window is open, but it won't stay open long.
Your membership also unlocks:
