Navigating Neurodivergence with AI Chatbots: Benefits, Tensions, and Implications for HCI

Paper Title

Navigating Neurodivergence with AI Chatbots: Benefits, Tensions, and Implications for HCI

Publication Info

• Topic area: Human-Computer Interaction (HCI) and neurodivergent users' experiences with AI chatbots.
• Keywords: Neurodivergence, AI chatbots, masking, emotional regulation, working memory, self-motivation, authenticity, privacy, social relationships, assistive technologies.

Background and Problem

• Problem / challenge: Existing research has not sufficiently examined how AI chatbots impact neurodivergent users' sense of privacy, authenticity, and social relationships. Prior studies highlight benefits but also limitations, such as generic mental health advice, biased communication strategies, and risks to social connections.
• Significance: Understanding these dynamics is critical for designing AI systems that better support neurodivergent users without reinforcing stereotypes or compromising their social and emotional wellbeing.
• Motivation and related work: Previous HCI studies have explored AI's role in mental health, companionship, communication, and productivity for neurodivergent users, but gaps remain in addressing tensions like masking and privacy. This paper builds on prior findings by conducting a qualitative study to explore these issues in depth.

Solution

• Proposed approach: A qualitative study based on 23 semi-structured interviews with neurodivergent participants to investigate their use of AI chatbots, motivations, and concerns.
• Novelty:
  1. Empirical investigation into condition-specific challenges driving neurodivergent users' reliance on AI chatbots.
  2. Identification of key tensions around masking, privacy, and social relationships in AI-mediated interactions.
  3. Insights into how AI chatbots simultaneously serve as accessible support and sources of risk for neurodivergent users.
• Procedure and key techniques:
  • Recruitment of 23 neurodivergent participants via university mailing lists and Reddit forums.
  • Semi-structured interviews conducted over Zoom, averaging 50 minutes.
  • Inductive and deductive multi-cycle coding of transcripts to identify themes, resulting in three primary codes: condition-specific challenges, use cases, and tensions with AI.

Results

• Concrete findings:
  • Participants used AI chatbots for therapy, communication, education, work, and task planning, driven by challenges like emotional regulation, working memory, and self-motivation.
  • Tensions included AI promoting masking, privacy concerns, and its impact on social relationships.
  • AI was perceived as a non-judgmental space but raised concerns about robotic communication and authenticity.
• Advantage over baselines: AI chatbots were valued for their constant availability, ability to summarize information, and support for emotional expression, but concerns about generic advice and biased outputs persisted.
• Experiments / evaluation: Interviews revealed diverse use cases (e.g., summarizing emotionally charged messages, planning tasks, resolving social conflicts) and highlighted tensions such as masking in communication and reliance on AI over human interaction.
• Limitations and future work:
  • Limited participant diversity (primarily younger, tech-literate adults).
  • Findings may lose relevance as AI technologies evolve.
  • Future research should include broader demographics and longitudinal studies to explore long-term impacts of AI on neurodivergent users.

Summary

This paper explores how neurodivergent people use AI chatbots to address condition-specific challenges like emotional regulation, working memory, and self-motivation. Through 23 interviews, it identifies benefits such as accessible support and task planning, alongside tensions around masking, privacy, and authenticity. The findings suggest that AI design should account for societal factors, including neurodivergent users' needs for authenticity and privacy. Future research should broaden participant diversity and examine long-term implications of AI use in neurodivergent communities.