Communications of the ACM, March 2019, Vol. 62 No. 3, Pages 54-60
Contributed articles: The Seven Tools of Causal Inference, with Reflections on Machine Learning
By Judea Pearl
“Unlike the rules of geometry, mechanics, optics, or probabilities, the rules of cause and effect have been denied the benefits of mathematical analysis.
…the art of automated reasoning.”
The dramatic success in machine learning has led to an explosion of artificial intelligence (AI) applications and increasing expectations for autonomous systems that exhibit human-level intelligence. These expectations have, however, met with fundamental obstacles that cut across many application areas. One such obstacle is adaptability, or robustness. Machine learning researchers have noted current systems lack the ability to recognize or react to new circumstances they have not been specifically programmed or trained for. Intensive theoretical and experimental efforts toward “transfer learning,” “domain adaptation,” and “lifelong learning” are reflective of this obstacle.
Another obstacle is “explainability,” or that “machine learning models remain mostly black boxes” unable to explain the reasons behind their predictions or recommendations, thus eroding users’ trust and impeding diagnosis and repair; see Hutson and Marcus. A third obstacle concerns the lack of understanding of cause-effect connections. This hallmark of human cognition is, in my view, a necessary (though not sufficient) ingredient for achieving human-level intelligence. This ingredient should allow computer systems to choreograph a parsimonious and modular representation of their environment, interrogate that representation, distort it through acts of imagination, and finally answer “What if?” kinds of questions. Examples include interventional questions: “What if I make it happen?” and retrospective or explanatory questions: “What if I had acted differently?” or “What if my flight had not been late?” Such questions cannot be articulated, let alone answered by systems that operate in purely statistical mode, as do most learning machines today. In this article, I show that all three obstacles can be overcome using causal modeling tools, in particular, causal diagrams and their associated logic. Central to the development of these tools are advances in graphical and structural models that have made counterfactuals computationally manageable and thus rendered causal reasoning a viable component in support of strong AI.
About the Author:
Judea Pearl is a professor of computer science and statistics and director of the Cognitive Systems Laboratory at the University of California, Los Angeles, USA.
