11. Bias in machine learning as a multi-dimensional problem

Nowadays, when talking about Artificial Intelligence (AI), we mostly refer to methods of Machine Learning (ML). Whereas these technologies have a huge potential, there are also risks and limitations, including the reflection of societal biases. This chapter discusses the term bias from different perspectives and argues that bias in machine learning is a complex multidimensional problem. The current limitations of technical solutions to this problem thus bring important implications for SSH researchers to consider.