Increasing Trust in AI with Explainable Artificial Intelligence (XAI): A Literature Review

Abstract

Artificial Intelligence (AI) is one of the most versatile technologies ever to exist so far. Its application spans as wide as the mind can imagine: science, art, medicine, business, law, education, and more. Although very advanced, AI lacks one key aspect that makes its contribution to specific fields often limited, which is transparency. As it grows in complexity, the programming of AI is becoming too complex to comprehend, thus making its process a “black box” in which humans cannot trace how the result came about. This lack of transparency makes AI not auditable, unaccountable, and untrustworthy. With the development of XAI, AI can now play a more significant role in regulated and complex domains. For example, XAI improves risk assessment in finance by making credit evaluation transparent. An essential application of XAI is in medicine, where more clarity of decision-making increases reliability and accountability in diagnosis tools. Explainable Artificial Intelligence (XAI) bridges this gap. It is an approach that makes the process of AI algorithms comprehensible for people. Explainable Artificial Intelligence (XAI) is the bridge that closes this gap. It is a method that unveils the process behind AI algorithms comprehensibly to humans. This allows institutions to be more responsible in developing AI and for stakeholders to put more trust in AI. Owing to the development of XAI, the technology can now further its contributions in legally regulated and deeply profound fields.

Ähnliche Arbeiten