To illustrate the need for causal reasoning in artificial intelligence, Judea Pearl uses the example of diagnosing malaria. He argues that instead of merely identifying a statistical correlation between fever and malaria, machines must be able to understand the causal relationship—specifically, that malaria causes fever. Establishing this causal framework enables machines to ask counterfactual questions, which allows for the automated modification of the framework in response to specific interventions.

Example of Causal Reasoning in AI: Malaria and Fever

We need to think about how truly intelligent machines would think. 

The KEY is to replace reasoning by association (Bayesian Network) with causal reasoning. 

University of Michigan - Ann Arbor

Judea Pearl believes current artificial intelligence research is stuck in probabilistic associations. He argues that as of 2018, AI is merely a more powerful version of what machines could already do three decades earlier, defining these machines as systems for "finding hidden regularities in large sets of data." He is unimpressed by deep learning's achievements, viewing them as amounting to little more than intensive curve fitting.

Learn Before

Related

Learn After