A Narrative Exploration of Improper AI Design & Execution and Possible Effects on Human Values

Abstract

Artificial Intelligence (AI) technology is increasingly being adopted in practice for all kinds of tasks. Nowadays, many examples exist of AI being improperly designed or used. This often leads to the violation of human values, e.g., equality or safety. Organizations can properly ground these human values in AI by composing principles. This paper focuses on providing examples of improper design and usage of AI from practice, with regards to a selection of human values. In total, 54 examples were identified using a narrative literature search technique regarding seven human values selected. In conclusion, the examples show that the problems regarding the design and execution can be interrelated but do not have to be. Also, improper design and execution of AI solutions do not have to be intentional by definition due to a lack of understanding of AI technology and implementation. Lastly, examples of principles are provided to ground human values.