Why Frontier Models Fail as Buy-Side Analysts - Janie Gu, Google DeepMind & Suyeol Yun, LinqAlpha

Why watch

Janie Gu from Google DeepMind and Suyeol Yun from LinqAlpha explain why frontier AI models fail at buy-side analyst tasks. The video details how reproducing 65 industry benchmarks revealed critical flaws including data contamination, score manipulation, and inconsistent results. They present MandateBench, a specialized benchmark testing whether models can translate a fund's mandate into a complete investment memo, grading reasoning quality rather than just final answers. This is essential viewing for ML engineers, AI product managers, and fintech developers building or evaluating financial AI systems.