In search of dark patterns in chatbots

• Autores: Traubinger, Verena; Heil, Sebastian; Grigera, Julián; Garrido, Alejandra; Gaedke, Martin
• Año de publicación: 2024
• Idioma: inglés
• Tipo de recurso: documento de conferencia
• Materia: Ciencias de la Computación e Información; Dark Patterns; Deceptive Design; Usability Smells; Conversational User Interfaces; Chatbots
• Institución: Comisión de Investigaciones Científicas de la Provincia de Buenos Aires

While Dark Patterns are widely present in graphical user interfaces, in this research we set out to find out whether they are also starting to appear in Chatbots. Dark Patterns are intentionally deceptive designs that trick users into acting contrary to their intention - and in favor of the organization that implements them. Chatbots, as a kind of conversational user interface, can potentially also suffer from Dark Patterns or other poor interaction design, sometimes referred to as Usability Smells. This keeps users from easily achieving their goals and can lead to frustration or limitations for users. To find Dark Patterns and Usability Smells, we analyzed user reports of negative experiences. Since we found no well known dataset of reports, we created the ChIPS dataset with 69 complaints from different web sources, and then classified them as one of 16 established Dark Patterns, potential new Dark Patterns, Usability Smells, or neither. Results show that, even though there are instances of established Dark Patterns, negative experiences usually are caused by chatbot defects, high expectations from users, or non-intuitive interactions.