The Need for Explainable AI

In the past few years, artificial intelligence (AI) has achieved impressive feats, much more than it ever did in the preceding decades. If we consider what happened in the last pandemic, AI was instrumental in helping the pharmaceutical industry develop a vaccine in record time. But AI’s extensive use is widespread in detecting all sorts of cancers, developing new drugs, and rehabilitating patients. On another front, we now have self-driving cars taking us from one place to the other, our airplanes are flown by an auto-pilot 90% of the time, and the level of safety we’ve reached is unheard of. If we look at basic human tasks, the ImageNet competition showed that even when it comes to simply labeling images (something we learned to do when we were still infants), an AI is capable of achieving much more accurate results. The list of examples can keep on going forever since AI systems can be found in most applications. The truth is that AI can...