Evals, Enterprise Challenges, and AI Future

Scale AI CEO on Meta’s $14B deal, scaling Uber Eats to $80B, & what frontier labs are building next

Topic 6 of 7

Published Oct 9, 2025 · Clip 28:42–35:27 (6:45)

Evals benchmark 'what good looks like' for systems, especially in enterprises. Jason explains probabilistic judgments over binary correctness. Improvements combine compute, models, and data. Future: From knowing to doing via agents in environments like Salesforce.