The measurement of attitudes towards Artificial Intelligence: An overview and recommendations

Abstract

The growth in use of Artificial Intelligence is having a major impact on society, with further impacts anticipated in the coming years and decades. There are individual differences in attitudes towards Artificial Intelligence and it is important for scientists and others to be able to measure these. Individual differences in attitudes towards Artificial Intelligence may be associated with other major psychological or circumstantial factors, and understanding these associations is beneficial. In addition, it is important to be able to track attitudes towards Artificial Intelligence over time. For this purpose, scientists have developed psychometric measurement tools to measure attitudes towards Artificial Intelligence. This Chapter provides an overview and evaluation of these tools, with a focus on tools that measure general attitudes towards Artificial Intelligence, and that are quantitative measurements, which can be analysed statistically. Semantic, methodological, and psychometric factors that the user should consider when choosing a suitable tool are discussed. The choice of measurement tool may depend on many researcher-driven considerations, including time, cost, and practical factors, but the quality and validity of the measurement tool should be a major factor in this choice. A scale’s ability to capture important dimensions in the data should also be a key consideration. We recommend that observed ambivalence about AI is best captured with a bi-dimensional AI attitudes scale.