AI definitions: Causal Inference
/Causal Inference - The scientific method for determining the cause-and-effect relationship between variables. Causal AI is the software application of that science. Getting to an exact cause can be difficult. A 2021 study found that even in reputable medical journals, a quarter of the published papers failed to identify the correct cause. This is one reason why an AI model can have a high degree of accuracy and still make poor recommendations. If a model is determined to be “accurate,” it means the AI is effective at identifying patterns. However, “accuracy” provides no information about whether those patterns will continue during intervention. In other words, is it possible for machine learning to make a good prediction, but not identify the cause accurately. Note: Most machine learning applications work fine without causal reasoning and do not need that added layer of engineering. It’s when the AI moves from pattern recognition to decision-making that causal reasoning can become essential.