How Wall Street is really using AI — from staff training to performance reviews (www.businessinsider.com)

Wall Street’s biggest banks are aggressively embedding generative AI across front-, middle- and back-office functions, reshaping workflows, roles and culture rather than running isolated pilots. Business Insider reports JPMorgan has deployed its in‑house genAI to hundreds of thousands of employees (part of an $18B tech budget) and CEO Jamie Dimon says a $2B AI push has already paid for itself in savings. Citi gives nearly 180,000 staff access to proprietary tools (≈7 million uses this year), reports saving 100,000 developer hours per week via automated code reviews, and is piloting agentic AI with 5,000 colleagues. Goldman, Morgan Stanley and Bank of America have rolled out company-wide assistants and productivity tools; Morgan Stanley’s DevGen.AI saved ~280,000 developer hours in six months, and Bank of America’s Erica handled 2 million customer interactions in a day. Technically, banks are combining model-driven assistants, agentic/multi‑step automation, and heavy data engineering—recognizing that AI value depends on large, well-governed data pipelines and integration with existing systems. Implications include big productivity multipliers for software engineers, faster client workflows, and systematic embedding of AI across processes, but also new risks: AI-enabled cyberattacks, governance and compliance headaches, and pressure to demonstrate ROI. Consulting firm estimates suggest AI could redefine roughly 44% of banking work by 2030, making robust model-, data- and risk-management strategies a strategic imperative.