The AI Black Box Issue And Explainable AI

        As the world evolves very fast, same goes to the tech industries, the world evolution works in hand with the advancement in technology. If technology hasn’t been upgrading, the world would pretty much be the same as it was 200 years back with little changes.

Explainable AI according to Wikipedia “refers to the methods and techniques used in the application of artificial intelligence technology, such that the results of the solution can be easily understood by humans”. It is an emerging field of research whereby individuals try to understand why and how certain machine learning models and algorithms arrive at a particular result or make decisions. It deals with machine learning models or artificial intelligence systems that produce results that humans can readily understand. It is the exact opposite of the “black box artificial intelligence” model which basically describes the methods as not interpretable.

The AI Black Box Issue

The need for explainable AI has become inherent especially in fields where the potential outcome or consequence is very important. Just like Dave Costenaro, Head of artificial intelligence said “If algorithm results are low-impact enough like the songs recommended by a music service, society doesn’t need regulators plumbing the depths of how these recommendations are made”.

Taking the words of the expert, we could actually live with an app misunderstanding or error in recommending songs for its users which is in no doubt a minor mistake, but blunders in certain cases whereby recommendations by an the system could bring huge damages particularly the healthcare industry cannot be discarded easily.

       As the black box can’t be understood even by some of its inventors it is regarded as a very risky for usage, especially in the health sector, different big companies Amazon have decided to apply the use of it even when the black box isn’t fully understood on things like job screening and others, the application of this has not been fully successful with well-identified problems like picking another gender or race over the other due to common historical factors.