Causal Machine Learning Initiative at CMU
The Causal Machine Learning Initiative at Carnegie Mellon University brings together the foremost researchers in the AI and machine learning space to develop better learning tools for AI.
The power of AI and machine learning algorithms have grown exponentially over the last 20 years, but one extremely important limitation involves causal knowledge. At Carnegie Mellon, we are taking AI learning to the next level. Causal learning helps answer critical why and what-if questions — key to making better decisions under uncertain circumstances. Unlike traditional AI, which finds patterns, causal AI understands cause and effect, allowing for smarter intervientions.
How Causal Learning Works
- Goes beyond correlation to support counterfactual thinking, identifying real causes, not just patterns.
- Uncovers hidden factors in order to find variables that impact decision-making.
- Creates reliable AI by enabling systems to reason and adapt like humans.
Why Causal Learning Matters
- Improves outcomes by helping us understand the hidden factors affecting success.
- Enhances decision-making by allowing us to predict impact before we act.
- Optimizes strategy and operations by making data-driven adjustments in complex environments.
Why is Carnegie Mellon the Home for Causal Machine Learning?
Machine learning-based techniques for acquiring causal knowledge from data — in particular non-experimental data — have exploded in power and sophistication over the last 25 years. These techniques have contributed to identifying the genetic drivers of some cancers, understanding the causal mechanisms that underlie learning fractions and diagnosing autism. They also have been applied to radically improve the efficiency of image refinement in generative AI.
Carnegie Mellon is home to some of the foremost interdisciplinary researchers in the AI and machine learning spaces in the world, in a wide array of departments and disciplines. The Causal Machine Learning Initiative will bring these researchers together in order to develop causal learning tools that will allow AI platforms or implementations of AI platforms to learn and incorporate causal knowledge into AI products.
The initiative will provide reliable platforms to discover causality, including hidden causal factors and causal influences, from observational or experimental data, and facilitate the integration of causal knowledge into AI implementations
More About Causal Learning at Carnegie Mellon
CMU CLeaR Group
Pioneering research in causal discovery, the CLeaR Group at CMU has developed foundational methods for uncovering causal relationships from data. Their work integrates insights from statistics, machine learning and philosophy to improve decision-making in complex environments.
TETRAD Initiative
A powerful open-source platform that allows researchers and practitioners to model and test causal relationships using real-world data, TETRAD provides tools for causal discovery, structural equation modeling and advanced statistical inference, enabling more precise and actionable insights.
For More Information
Richard Scheines
Bess Family Dean, Dietrich College of Humanities and Social Sciences
scheines@cmu.edu
Adam Causgrove
Senior Director, Strategic Partnerships
causgrove@cmu.edu
Peter Spirtes
Mariana Brown Dietrich Professor and Head of the Department of Philosophy
ps7z@cmu.edu
Kun Zhang
Associate Professor of Philosophy
kunz1@cmu.edu