Next Article
AI models ace CFA Level III exam, human touch still vital
Technology
A new study from GoodFin and NYU Stern shows advanced AI models can now pass the notoriously tough CFA Level III exam—a big deal in the finance world.
Out of 23 models tested, OpenAI's o4-mini led with a 79.1% score, followed by Google's Gemini 2.5 Flash at 77.3% and Anthropic's Claude Opus 4 at 74.9%.
These AIs handled both tricky essay questions and other exam components without special training.
While AI is taking over number-crunching, human analysts are still needed
Even with these AI wins, CFA experts say human analysts are still key for things like understanding clients and showing empathy—areas where AI just can't compete yet.
They see AI taking over routine number-crunching, letting real advisors focus more on personal advice and building trust with clients.