Industry2h ago
Bankers Find No AI Outputs Client-Ready
Zero AI outputs from models like GPT-5.4 and Claude Opus 4.6 passed a review by 500 investment bankers. The results proved too imprecise or incorrect for direct delivery. Despite the failure, over half of the participants will use these outputs as initial drafts. This confirms a persistent gap between LLM capabilities and high-stakes financial precision.