Understanding responsible artificial intelligence : a case study on the considerations to made and how they can be addressed

The aim of this thesis is to contribute with new insights on the concept of responsible artificial intelligence (RAI), by answering the following main research question: How can we understand responsible artificial intelligence? We stand at the precipice of a new era with rapid advancements in artificial intelligence (AI). Though AI is already deeply embedded in our society and almost every industry, companies might not know how to take a responsible approach to AI. The area of RAI has gained limited attention in academia and little research has been conducted on the concept. The purpose of our master thesis has therefore been to shed light on the concept of RAI, including which considerations that should be made and how these can be addressed when working toward RAI. To do so, we have conducted a single case study on Equinor and collected qualitative data through semi-structured interviews with the employees. We find that RAI means to take a thorough and holistic approach to how one can use AI responsibly, it entails acknowledging the importance of humans when using AI, and it demands an understanding of both responsibility and AI. This understanding of RAI can be expressed in two main findings; (i) Humans are more important than expected and (ii) understanding responsibility and AI is a prerequisite. First, acknowledging the importance of humans when using AI involves holding humans responsible for the AI, entrusting humans to ensure that ethical principles are maintained, placing humans in control of AI, utilizing the knowledge and experience of the employees rather than simply replacing them with AI, and designing the AI in a way that facilitates humans doing what they do best and being able to fulfill their responsibilities. Second, an understanding of responsibility that facilitates RAI, is the notion that responsibility entails doing more than what is required or expected. The need for an understanding of AI is based on the ability it creates to mitigate the possible negative outcomes of AI and ensure transparency, and thereby trust and acceptance of AI. This understanding is also at the core of an RAI strategy. Based on our findings, we believe that when a company understands and acts in accordance with these insights, it has achieved Responsible Artificial Intelligence.nhhma