Nearly three in four consumers using AI for banking tasks opt for ChatGPT, Claude or Gemini — almost double those using a bank’s chatbot, a McKinsey report found.
As banks seek to strike a balance between accelerating AI use and maintaining operations and security, they’re also contending with digital technology challenges.
Thousands of customers from JPMorgan Chase, among other firms, reported issues accessing their online accounts Monday.
While navigating traditional technology challenges, data readiness continues to be an issue for chief information officers on their AI journeys.
Nearly half of banking executives ranked data readiness as their top AI obstacle, McKinsey found. To see better results with AI projects over time, banks need to build institutional intelligence – starting with data that agents can consume at scale, according to McKinsey.
Banks also need to build context, which McKinsey described as a map of customer accounts with permissions set in place, and implement an encoded judgment call policy agents can reuse.
“The customer wants a capable deputy that asks for permission, and the bank that gets the permission model right may win the mandate to move the customer’s money,” the report said.
Still, high-value AI deployments remain stuck in pilot purgatory, largely due to meaningful business data also being trapped inside siloed legacy systems, according to a report from information management compan y Iron Mountain. Instead of spending time scaling agents, skilled IT and data science personnel are burning hours wrangling data manually to support projects, according to 84% of IT leaders.
Source link







