The Behavioral Design Podcast · Samuel Salzer and Aline Holzwarth

Chatbots, Synthetic Users and AI for User Research with Hassan Aleem

·41 min·2 clips

Hassan says synthetic users are useful only as a validation tool, not a replacement for real research.

1. The Behavioral Design Podcast episode centers on Hassan Aleem and AI for user research, chatbots, synthetic users, and literature reviews. 2. Samuel hosts Hassan Aleem, a Nuance Behavior colleague with a PhD in neuroscience and experience in neuroaesthetics, fintech, health wearables, and public health. 3. The episode asks what behavioral science should do as AI moves to the front of product design and research. 4. Hassan says AI is strongest at making tasks easy, conversational, and low-friction through tools like ChatGPT voice mode. 5. He argues that making something easy is not the same as driving motivation, engagement, or long-term behavior change. 6. He uses Duolingo and language learning to show that even very smooth AI interactions still need structure and rewards. 7. Samuel describes testing advanced voice mode with his mother as a personal user experiment. 8. Hassan points to Duolingo’s gamified design as evidence that long-term learning needs more than a conversational interface. 9. He says behavioral science matters because domain expertise has to shape what AI tools are actually good for. 10. He describes a sustainability NGO chatbot project that answers recycling and waste questions. 11. That project includes segmentation between highly motivated waste-conscious users and people who only want a fast answer. 12. Hassan says the team is building QR-code flows for object-specific questions, such as whether a water bottle should be recycled. 13. He argues that some situations need a yes-or-no answer rather than a long chat. 14. He warns against a “form factor trap,” where teams choose chat because chat is trendy. 15. He says customer service systems can become annoying when they force fake empathy before giving basic information. 16. On synthetic users, he moves from initial skepticism toward cautious optimism. 17. He compares synthetic-user modeling with physics-style simulation, but says human psychology is not understood with that level of control. 18. He says synthetic users are best used for validation, perspective taking, and cross-checking real studies rather than replacing them. 19. The conversation is interview-led, practical, and examples-driven, with Samuel pressing for concrete product and research implications. 20. Behavioral scientists, product researchers, and UX teams will get the most value here; listeners wanting a purely conceptual AI debate may skip it.

Listen to the show on

Apple Podcasts Spotify

Play clips from this episode More from The Behavioral Design Podcast