Chatbots and Procrastination : Identifying and Mitigating Risks

Sammanfattning

Procrastination is widespread and closely tied to well-being, stress, and academic and work performance. This explorative study investigates potential risks of using large language model chatbots to treat procrastination, as well as design approaches that could mitigate these risks. Experts in three areas, psychology, language technology, and human-computer interaction, were interviewed, and thematic analysis was conducted on the resulting material. Five themes emerged, concerning individual adaptation, engagement and motivation, alignment and feedback, social factors, and anthropomorphism. Key risks include alignment conflicts between general and specialised chatbots, loss of social accountability, anthropomorphism-induced overtrust, and over-generalisation in treatment. Proposed design responses include gradual user modelling, predefined personality presets, simulated accountability mechanisms, language adaptation, and redirection to alternative treatments when appropriate. The study contributes to understanding the specific challenges of using chatbots for procrastination reduction and suggests that while some risks are inherent to the technology, others are addressable through design and warrant consideration by developers.