#358 How AI Agents Will Work While You Sleep | Ruslan Salakhutdinov, Professor at Carnegie Mellon

About this episode

The episode explores the reliability and use cases of AI agents, featuring insights from Ruslan Salakhutdinov.

Almost every AI agent demo lands in roughly the same place: it works most of the time, looks remarkable, and then fails in a way no one anticipated. Self-driving cars hit this wall a decade ago, and agents are running into it now. For data and AI teams, the question is no longer whether agents can complete a task — it's whether they can complete it reliably enough to remove the human reviewer. Which categories of work tolerate a 90% success rate? Which absolutely don't? And where should the next layer of guardrails sit? Ruslan Salakhutdinov is a UPMC Professor of Computer Science at Carnegie Mellon University and one of Geoffrey Hinton's former PhD students. He has previously served as Director of AI Research at Apple and VP of Research in Generative AI at Meta. His research focuses on deep learning, reasoning, and AI agents. In the episode, Richie and Russ explore the most exciting use cases of AI agents today, long horizon tasks, the credit assignment problem, multi-agent systems, designing reliable human-in-the-loop workflows, agent safety and guardrails, embodied and physical AI, lessons from self-driving cars, the difference between academia and industry, and much more…

People in this episode

Host: Richie

Guest: Ruslan Salakhutdinov

Topics covered

• AI agents
• deep learning
• human-in-the-loop workflows
• agent safety
• self-driving cars
• multi-agent systems

Keywords

• AI agents
• reliability
• self-driving cars
• deep learning
• human-in-the-loop
• multi-agent systems
• guardrails

Mentioned in this episode

Organizations: Carnegie Mellon University, Apple, Meta, Anthropic, Waymo

Books & works: DeepSeek-V3 Technical Report, Kimi K2 Technical Report

More episodes of DataFramed

• #362 How to Have a Machine Learning Career in 2026 | Marina Wyss, Senior Applied Scientist at Twitch · June 1, 2026 · 48 min
• #361 If You Want AI to Work, Fix This Boring Thing First with Veronika Durgin, VP of Data at Saks · May 25, 2026 · 49 min
• #360 What's Your Biggest AI Ethical Nightmare? | Reid Blackman, CEO at Virtue Consultants · May 18, 2026 · 57 min
• #359 My Best Friend is AI with Valerie Tiberius, Professor of Philosophy at University of Minnesota · May 12, 2026 · 44 min
• #357 Data-Driven Workforce Analytics with Ben Zweig, CEO at Revelio Labs · April 27, 2026 · 58 min
• #356 The Forecast for Time Series Forecasts with Rami Krispin, Senior Manager of Data Science at Apple · April 20, 2026 · 54 min