AI's Fluency Is Not Understanding | Elias Bareinboim S3E02 | CausalBanditsPodcast.com

Send us Fan Mail

This episode was created in collaboration with the UBC Causal AI Cluster: https://aicausal.ubc.ca/

Ask an image model what happens when you flip the switch on a desk lamp, and many will show it lit, even when the cable is lying on the desk, unplugged.

The lamp is only the easiest instance. The same thing happens wherever the question is about an action or a change the data never showed, which is why fixing one example does not fix the problem. Elias Bareinboim argues that the failure is structural: these systems are built to exploit correlational signals, not to understand the mechanics of the world that produced their training data. The next generation of AI, he argues, needs a causal model of the world and the reasoning machinery to use it.

Will we need a breakthrough to move AI to the next level? And how can today's models be so fluent while failing at questions of cause and effect?

In this episode, we cover:

* Where are we with the AI systems of today?
* The difference between fluency and understanding
* The five capabilities a causally intelligent system needs
* Why Elias thinks the future of AI has to be causal, and what it will cost us if we only scale

Enjoy the episode!

Recorded on Sep 23, 2026 in New York City, US.

About The Guest
Elias Bareinboim is a professor of computer science at Columbia University, where he directs the Causal Artificial Intelligence Lab, editor-in-chief of the Journal of Causal Inference, and founder of CausalX, a new lab building the causal layer that modern AI is missing.

He earned his PhD under Judea Pearl at UCLA and is one of the leading researchers in causal AI and causal reinforcement learning, a field he helped establish. His book, Causal Artificial Intelligence: A Roadmap for Building Causally Intelligent Systems, is forthcoming from MIT Press, with a free draft available online.

Connect with Elias:

• Elias on LinkedIn: linkedin.com/in/eliasb
• Elias on the Internet: https://causalai.net/

About The Host
Aleksander (Alex) Molak is an independent machine learning researcher, educator, entrepreneur and a best-selling author in the area of causality (https://amzn.to/3QhsRz4 ).

Connect with Alex:

• Alex on the Internet: https://bit.ly/aleksander-molak

Links

• Bareinboim, E. - "Causal Artificial Intelligence: A Roadmap for Building Causally Intelligent Systems" (forthcoming, MIT Press), free draft (https://causalai-book.net/)
• Bareinboim, Correa, Ibeling, Icard - "On Pearl's Hierarchy and the Foundations of Causal Inference" (R-60) (https://causalai.net/r60.pdf)
• Bareinboim, E. - "An Introduction to Causal Reinforcement Learning" (R-65) (https://causalai.net/r65.pdf)
• Bareinboim, Forney, Pearl (2015) - "Bandits with Unobserved Confounders: A Causal Approach" (https://papers.nips.cc/paper_files/paper/2015/file/795c7a7a5ec6b460ec00c5841019b9e9-Paper.pdf)
• Pearl, J. & Mackenzie, D. - "The Book of Why" (https://amzn.to/4e0ATrZ)

Resources
💠 Join FREE Causal Python Bi-Weekly Newsletter: https://causalpython.io
💠 Alex's Book on Causality: https://amzn.to/3SKRXIw
💠 Get notifications about new episodes: https://causalbanditspodcast.com

Let's connect!
🔷 Linkedin: https://www.linkedin.com/in/aleksandermolak/
🔷 Bluesky: https://alxndrmlk.bsky.social
🔷 Tiktok: https://www.tiktok.com/@alex.molak

Business
🔷 Consulting and Causal AI Training For Your Team: [email protected]

Podcast Playlist
https://www.youtube.com/playlist?list=PLhKKv6iMja4p5FbJIgzTOE67E1M6c8lnB

#machinelearning #causalai #causalinference #causality

Support the show

Causal Bandits Podcast
Causal AI || Causal Machine Learning || Causal Inference & Discovery
Web: https://causalbanditspodcast.com

Connect on LinkedIn:   https://www.linkedin.com/in/aleksandermolak/
Join Causal Python Weekly: https://causalpython.io 
The Causal Book: https://amzn.to/3QhsRz4