From knowing too little to knowing too much Automated Emotion Recognition and Implications for Autonomous Agency

Abstract

Abstract: If one accepts that affect has an important role to play in reasoning and deliberation, the ability to accurately recognise and predict emotions and associated behaviour by Artificial Intelligence Systems (AIs) allows for cases of manipulation and restriction of agency. To support this claim, I give a brief discussion on cases of ‘knowing too little’ where AIs are tasked with making predictions on the agents’ emotions without having enough information to guarantee that these predictions would be accurate. As with other issues in AI research, such issues may dissolve with the advance of technological progress. In the main part of the chapter, I deal with issues arising from AIs being able to accurately recognise human emotion and predict actions associated with those emotions. I call these cases of ‘knowing too much’ to highlight issues relating to manipulation. I show that in these cases, there is a restriction of agency such that it becomes difficult to claim that the agent is autonomous or that they can be held responsible for any resulting action or behaviour. I conclude that solving the problem of ‘knowing too little’ will inadvertently lead to the creation of the second problem and suggest some important questions for further research.