Data on our Minds: Affective Computing at Work

We identify a new frontier in the rapid development and applications of inferential biometrics-based systems at work. This frontier marks a shift in the scope, manner and bases for managerial decision-making, as wide-ranging, automated inferences, classifications and measurements relating to intimate details about who a person is and how they are likely to behave become more significant. We call this ‘Algorithmic Affect Management’ (AAM). We find that AAM has significant regulatory implications across domains, from definitions to the need for specific, updated protection in both hard and soft law. We find that such a multi-layered, systems approach, covering both individual and group impacts, is required to adequately protect working people. Our research points to the need to mandate robust due diligence, auditing, monitoring and mitigation provisions on responsible actors who should report on significant risks and impacts – including from AAM at work – across supply chains and the AAM life cycle. In addition to clarifying the application of existing protection, we also recommend public consultation on the development of new, freestanding rights to explicitly protect against ‘neurosurveillance’ and the use of AAM to manipulate, interfere or commodify workers’ cognitive, emotional and behavioural functions and capabilities. We see this as an extension and application of fundamental human rights, labour and health and safety protection, with particular regard to privacy, human dignity and integrity of the person, in recognition that the new and increasingly intrusive sphere of AAM at work may demand a new category of protection. Regulators, the DRCF and the Fair Work Agency will need additional resources to consider and collaborate on governing AAM at work, investigate use cases, share insight, support AAM literacy and develop this dual approach.