Reliability: Understanding Cognitive Human Bias in Artificial Intelligence for National Security and Intelligence Analysis

• In Event: STS and Security Studies: Expertise, Infrastructures and Practices - III

Abstract

This paper seeks to contribute to science and technology studies through understanding cybersecurity of artificial intelligence, its connection to the intelligence community in the handling of massive amounts of data, and how much to trust intelligence analysis of artificial intelligence. Artificial intelligence is innovative in form and it is essential to acknowledge that human bias is a byproduct of evolution. As culture evolves by social, economic, political, and technological innovations, so too is the inclination towards primal and survival mode. Artificial intelligence that has skewed in its normalcy due to algorithmic mishaps intended to minimize human involvement and emotion have negatively transformed its potential. Understanding and acknowledging human bias in the data used by artificial intelligence helps the intelligence community with issues that are of critical interest to national security, such as cybercrimes, infectious diseases, and autonomous cars. The purpose of this paper is to open the artificial intelligence black box, understand how much to trust artificial intelligence analysis, and analyze the risks involved. This research is positioned in the cybersecurity and intelligence disciplines in combination with science and technology studies. Qualitative data collection of artificial intelligence such as congressional hearing reports and related documents will be gathered and analyzed using grounded theory. This research paper is an important contribution since artificial intelligence is operating at the speed of light and more data than we can humanly manage is accessible even for large intelligence agencies. Trustworthiness is a factor for mutual confidence of artificial intelligence and its technological innovation.

Author

• Gaudys Laxury Sanclemente, FLACSO