Open Data Science Conference

🎙️In this episode, we speak with Michael Littman, Brown University’s first Associate Provost for AI and a leading researcher in reinforcement learning.

The conversation builds on Michael’s ODSC AI East Keynote, “Conveying Tasks to Computers and How Machine Learning Machinery Can Help,” and explores one of the central challenges in AI: how humans communicate intent to machines.

Michael explains why large language models have made it easier for people to create software, automate tasks, and work with AI systems, but also why this shift makes human judgment more important. The discussion covers the limits of English as a programming language, how AI systems learn the wrong thing, the role of reinforcement learning and world models, and what universities and employers need to rethink as AI changes education and entry-level work.

🎧 Listen here: https://hubs.li/Q04k1TvY0

Key Topics Covered:
🔹Why communicating clearly with AI systems is harder than it looks
🔹What Michael’s Claude Code classroom experiment revealed about AI-assisted programming
🔹Why human judgment and validation matter when AI produces polished but imperfect work
🔹How programming, supervised learning, reinforcement learning, and inverse reinforcement learning help humans convey tasks to computers
🔹Why AI systems can optimize for the wrong goal, including the boat racing example
Michael’s perspective on world models, model-based reinforcement learning, and Yann LeCun’s argument
🔹How AI is changing higher education, assessment, and entry-level work
🔹Why adaptability, problem-solving, and lifelong learning remain essential skills

Memorable Outtakes:
💬Michael Littman on communicating with AI: “It can't read your mind. It has- you have to say it out loud.”
💬Michael Littman on AI-generated code: “You can always ask Claude code, it will always do something, but it won't always do the right thing, and being able to validate that it's actually done the right thing is extremely hard.”

References & Resources:
- Michael Littman – (https://hubs.li/Q04k22830)https://hubs.li/Q04k215C0
- Brown University AI – (https://hubs.li/Q04k1Zqz0)https://hubs.li/Q04k1MgN0
- Claude Code – https://hubs.li/Q04k21Tl0
- AlphaGo – https://hubs.li/Q04k1Xqk0
- Claude Design – https://hubs.li/Q04k1RsJ0
- AlphaGo – https://hubs.li/Q04k1WmP0
- OpenClaw – https://hubs.li/Q04k1R1w0

🎧 Listen here: https://hubs.li/Q04k1TvY0

The surge in OpenClaw’s popularity, gaining hundreds of thousands of GitHub stars in just months, signals a broader shift in how developers build AI: moving from passive question-and-answer bots toward proactive, autonomous agents.

How OpenClaw is Going Mainstream The surge in OpenClaw’s popularity, gaining thousands of GitHub stars in just months, signals a broader shift in how developers build AI.

After an incredible ODSC AI East, we’re bringing the energy, ideas, and hands-on learning to ODSC AI West 2026.

Join AI builders, data scientists, engineers, and leaders for three days of practical sessions, training, networking, and real conversations about what’s next in AI.

📍 San Francisco | Oct 27–29
👉 Register now: https://hubs.li/Q04k21Zr0

Discover how to measure AI ROI, identify the business value of AI projects, and move from experimentation to measurable impact across your organization.

The Velocity of Trust: How Governed Ex*****on Accelerates AI ROI AI ROI is becoming a top priority for business and technical leaders. Learn how to measure AI impact and turn AI projects into business value.

🚀 Experience the future of AI at ODSC AI West 2026

AI builders are moving from experimentation to ex*****on — and this year’s tracks are built around the skills, systems, and strategies shaping that shift.

Join us in San Francisco, October 27-29 to explore the future of AI across:
🤖 Physical AI
🧠 Agentic AI
⚙️ AI Engineering & AIOps
📊 Data Science & Machine Learning
☁️ Data & AI Infrastructure

From production agents and AI engineering workflows to scalable infrastructure and applied ML, ODSC AI West brings together the technical sessions, hands-on training, and industry insight AI teams need now.

⏳ Ends Friday: First 100 Passes

👉 Explore the tracks and secure your pass: https://hubs.li/Q04k1Skw0