Augmenting data analysis with human insight in AI

Abstract

AI has brought a significant shift in the analysis of data by automatically detecting patterns, predicting trends, and handling large amounts of data on a large scale in various domains like health care, finance, and security. Yet, completely independent AI systems face challenges like a lack of understanding of the context, being unclear about the process of functioning, and the inability to use ethical thinking. This emphasizes the significance of human ability in the process of AI. This article proposes a Human AI Augmented Data Analysis framework with the aim of developing a structured alliance between the intelligence of humans and the capabilities of AI systems. This framework is based on the three basic concepts of Human-in-the-Loop, Explain ability of AI Systems, and Collaborative Intelligence. This ensures the interaction between humans and AI systems at every step of the process of AI functioning with the four stages of the iterative process of Data Preparation, Model Development, Result Interpretation, and Decision Validation. This is unlike the majority of the existing frameworks, which function in a one-way or partial manner, as the proposed method is based on continuous interaction between human beings and AI. Both function together, rather than separately. The proposed framework is expected to overcome the following challenges. The system can be made transparent, the results can be accurate, and the chances of bias can be reduced. The proposed method can also generate trust in the system and can make human thinking and artificial intelligence close to each other while analyzing the data.