Causal inference is expensive. Here’s an algorithm for fixing that.
Does eating caviar make a person healthy – or even wealthy?
As humans, we intuitively understand that consuming fish eggs doesn’t confer well-being or riches. If anything, the opposite is true: Those who can afford to eat delicacies often do, while also having access to superior healthcare. But such logical leaps are generally beyond the capabilities of today’s narrow AI systems. Narrow AI struggles to differentiate between actions or states that appear in proximity (correlation) and actions that actually affect each other (causation).
As industry increasingly looks to AI to make decisions, the current methods of identifying co-variants will become obsolete. Whether treating patients or trouble-shooting manufacturing inefficiencies, it’s not enough to identify or even to predict symptoms. The path to curing both biological and mechanical breakdowns involves understanding the complex chain from instigation to outcome. We need to get to root causes.
Causal inference methods have made some progress toward this goal thanks to an improving ability to infer causal relationships from data. We’re pushing further. We’re building AI systems that enable operators to test for causes and identify paths to performance gains. We’re building AI systems that go beyond merely understanding that wealth leads to health and caviar consumption. They’ll advise how to obtain wealth and health in the first place.
Reverse-engineering causal graphs with soft interventions