About | HeinOnline Law Journal Library | HeinOnline Law Journal Library | HeinOnline

LOG IN
LOG IN


26 Bialostockie Studia Prawnicze 25 (2021)
Bias in Artificial Intelligence Systems

handle is hein.journals/bialspw26 and id is 385 raw text is: Bialystok Legal Studies Bialostockie Studia Prawnicze 2021 vol. 26 nr 3 S C e n d o DOI: 10.15290/bsp.2021.26.03.02 Received: 19.12.2020 Accepted: 25.03.2021 Rafal Rejmaniak University of Bialystok, Poland r.rejmaniak@uwb.edu.pl ORCID ID: https://orcid.org/0000-0003-1908-5844 Bias in Artificial Intelligence Systems Abstract: Artificial intelligence systems are currently deployed in many areas of human activity. Such systems are increasingly assigned tasks that involve taking decisions about people or predicting future behaviours. These decisions are commonly regarded as fairer and more objective than those taken by humans, as Al systems are thought to be resistant to such influences as emotions or subjective beliefs. In reality, using such a system does not guarantee either objectivity or fairness. This article describes the phenomenon of bias in Al systems and the role of humans in creating it. The analysis shows that Al systems, even if operating correctly from a technical standpoint, are not guaranteed to take decisions that are more objective than those of a human, but those systems can still be used to reduce social inequalities. Keywords: Al discrimination, Al fairness, algorithmic bias, artificial intelligence Introduction Technological solutions based on artificial intelligence (AI) are being used more and more widely in various spheres of human activity. AI systems are deployed in both the private and the public sectors. The widespread use of such solutions is motivated by the potential benefits, which are hard to overestimate - from making production processes more efficient or analysing large quantities of data at speeds far exceeding human capabilities to forecasting future events. One of the frequently cited properties of AI systems, said to give them an advantage over humans in performing certain types of tasks, is the greater objectivity of their 'decisions' and © Faculty of Law, University of Bialystok, Poland 25