Bank of America launched its AI-powered chatbot, Erica, in 2016. After years of upgrades and patents, it now manages about 2 million customer interactions daily—the workload of roughly 11,000 employees. However, this efficiency comes at a steep price: the bank has spent nearly $120 billion on technology since then, including $12 billion last year alone, with $4 billion for development and $8 billion for maintenance. High costs reflect the caution required when deploying tools like generative AI, where errors can damage trust and waste resources.
BofA’s tech budget surpasses the total expenses of many banks in the KBW index. Its consumer workforce has shrunk from 101,000 in 2011 to 55,000 due to automation, and fraud losses have halved since 2018. Alongside Capital One, it holds 65% of banking AI patents. Yet, most banks reveal little about AI returns—many report under 10%. Data preparation alone costs billions, as seen with BofA and Morgan Stanley. JPMorgan spends $2 billion annually on AI, achieving comparable savings. Despite potential gains, experts warn that AI could widen the gap between the largest banks and smaller rivals.
