Using Ethical Guidelines for Defining Critical Quality Requirements of AI Solutions

Artificial Intelligence (AI) has become a fast-growing trend and an increasing number of companies are interested in developing AI solutions. Many companies have realised that black box nature of AI technologies can raise ethical questions. And they are defining ethical guidelines which can be utilised when developing AI solutions. The goal of this thesis was to explore how companies can solve potential ethical issues of AI solutions during requirements engineering activities. The main research method used in this thesis was a multiple-case study approach. Semi-structured interviews and document analysis were carried out to discover the AI ethical guidelines in three Finnish companies. The results are based on the open coding of the interview data and the ethical guidelines of the case companies. The ethical guidelines developed by the case companies focus on solving potential ethical issues like accountability, fairness, privacy, safety, security, transparency and trust. All three case companies devised ethical guideline relating to fairness, transparency and trust. To capture different views on critical ethical issues, two of the case companies recommend using multi-disciplinary development teams consisting of experts like data scientists, designers, and sociologists. The results of this thesis suggest that companies can utilise the ethical guidelines of AI solutions to identify and prioritise the essential quality requirements of AI solutions