Gender Bias in Artificial Intelligence: The Need for Diversity and Gender Theory in Machine Learning

摘要

Artificial intelligence is increasingly influencing the opinions and behaviour of people in everyday life. However, the over-representation of men in the design of these technologies could quietly undo decades of advances in gender equality. Over centuries, humans developed critical theory to inform decisions and avoid basing them solely on personal experience. However, machine intelligence learns primarily from observing data that it is presented with. While a machine's ability to process large volumes of data may address this in part, if that data is laden with stereotypical concepts of gender, the resulting application of the technology will perpetuate this bias. While some recent studies sought to remove bias from learned algorithms they largely ignore decades of research on how gender ideology is embedded in language. Awareness of this re-search and incorporating it into approaches to machine learning from text would help prevent the generation of biased algorithms. Leading thinkers in the emerging field addressing bias in artificial intelligence are also primarily female, suggesting that those who are potentially affected by bias are more likely to see, understand and attempt to resolve it. Gender balance in machine learning is therefore crucial to prevent algorithms from perpetuating gender ideologies that disadvantage women.