MIT Class Integrates Anthropology to Redesign Humane Chatbots

A novel undergraduate course at the Massachusetts Institute of Technology (MIT) is bringing together the fields of anthropology and computer science to reimagine the design and purpose of AI chatbots. The course, known as Humane User Experience Design (6.S061/21A.S02), encourages students to think beyond traditional uses of chatbots as digital distractions and instead develop AI tools that function as social guides and moral partners for users, particularly young adults.

The course is the result of collaboration between Professor Arvind Satyanarayan, a computer scientist specialising in data visualisation and user interfaces, and Professor Graham Jones, an anthropologist with a focus on communication. Drawing on a grant from the MIT Morningside Academy for Design, their course is cross-listed between computer science and anthropology departments, offering students a rare opportunity to fulfil humanities requirements while honing technical skills relevant to their future careers.

Core to the curriculum is the integration of linguistic anthropology methods into chatbot development. Students are taught to address the interactional and interpersonal needs of humans, contrasting with the standard approaches in human-computer interaction, which often prioritise gathering data over fostering genuine connection.

"Human-computer interaction inherited methods from anthropology, like interviews and observation studies, but these have become diluted. We lose vital rapport with users this way," Satyanarayan noted. Jones added that the interdisciplinary approach helps students view human conversation as a rich source of design inspiration for AI systems.

Student projects highlight the impact of this approach. One team created Pond, a chatbot designed to support recent college graduates transitioning to independent adult life. Rather than simply echoing user sentiment or offering generic encouragement, Pond provides practical advice, teaches soft skills such as setting boundaries, and utilises interactive role-playing to foster confidence in real-world scenarios. Features include a point system for skill development and interactive prompts, demonstrating a blend of engaging user experience with evidence-based behavioural guidance.

Another project, News Nest, seeks to encourage engagement with credible news sources while avoiding common digital pitfalls such as 'doomscrolling.' The chatbot employs a unique approach: rather than using human avatars, it introduces various bird personas, each covering a different news domain. The system promotes media transparency by always displaying news sources and political leanings, building media literacy in a playful way.

A third team developed M^3 (Multi-Agent Murder Mystery), a social deduction game where users interact with multiple advanced chatbots—including Gemini, ChatGPT, Grok, and Claude—as fellow players in a murder mystery scenario. The game leverages distinct chatbot personalities and encourages critical thinking, showing that AI interactions can be educational and entertaining while avoiding excessive user attachment.

These projects underscore the benefits of incorporating the humanities into AI education, allowing students to create systems that better align with nuanced human needs. The course has also demonstrated real-world relevance: students report that their classroom experience directly mirrors industry practices, leading to concrete job opportunities.

MIT's interdisciplinary experiment points to a broader trend in AI development, where understanding human context and values can lead to more responsible and supportive technologies—an approach increasingly relevant as AI becomes embedded in daily life.

Source: news.mit.edu{:target="_blank"}