Chain of Thought | Intro to Scale's Agentic Leaderboards
Автор: Scale AI
Загружено: 2025-10-08
Просмотров: 85732
Описание:
In this episode of Chain of Thought, @Scale_AI 's Brad Kenstler (Head of Agent Capabilities and Environments) sits down with Bing Liu (Head of Research) and Chetan Rane (Head of Product, Agents) to discuss one of the most important frontiers in AI: Agentic AI.
They break down what “agents” really are, why they matter, and how the next generation of intelligent systems will be trained, evaluated, and deployed. From reinforcement learning environments to agent evaluation benchmarks, this conversation goes deep into the technical and research challenges shaping the future of AI agents.
Topics covered:
00:00 - What is Scale’s Research team focused on?
2:51 - How Scale's Product and Research teams collaborate to advance agentic AI
7:28 - Defining “agents” — and why environments are key to their performance
14:25 - Industry trends driving agent capabilities and autonomy
27:05 - Evaluating agents — current and next gen benchmarks
37:16 - How agents learn through RL and verifiers
50:59 The future of agents: capability merging, world models, and long-horizon planning
Learn more at http://bit.ly/4nJIwXr
Повторяем попытку...
Доступные форматы для скачивания:
Скачать видео
-
Информация по загрузке:
