New report: Ten themes for responsible and inclusive AI assistants – with a focus on neurodivergent youth

For many people, AI tools are simply a faster way to search for information or draft an email. For young people with cognitive and/or intellectual disabilities, they can mean much more: access to information that used to be out of reach, support that never runs out of patience, and a safe space to ask questions that feel too difficult to ask another person. At the same time, the risks are significant, including manipulation, misinformation and sensitive personal data ending up in the wrong hands. Possibilities and risks therefore need to be considered together, not in isolation.

The report is based on interviews with parents and guardians of young people aged 11 to 20 with neurodevelopmental conditions and/or intellectual disabilities, together with a review of current research. From this, we have identified ten themes for responsible and inclusive design of AI assistants:

  • Enable personalisation and adaptation
  • Support user autonomy and empowerment
  • Offer multimodal and accessible interaction design
  • Give non-judgmental support
  • Provide structural support and simplified guidance
  • Motivate and offer tireless support
  • Design for natural pause and reflection
  • Clarify that it’s a tool – like ‘a robot’
  • Bridge to real-world interaction and connection
  • Ethical safeguards and bias mitigation

Each theme concludes with concrete considerations for those designing, procuring or regulating AI tools. Much of what is needed for this particular group also makes AI assistants better for everyone, in line with the principle of universal design.

The report closes with a reflection on responsibility. When the same company both profits from engagement and data and is expected to protect its users, conflicts of interest are built into the system. As a thought model and a starting point for discussion, we therefore propose an ecosystem in which more actors take on responsibility – not instead of AI companies, but alongside them. After all, most of what is needed is not a matter of technical challenges, but of deliberate design choices.

The report is written by Jenny Lindström Beijar, Our Normal, and Anna Kolback, Studio Berget, as part of the EU co-funded project European Citiverses Uniting for Inclusiveness.

Want to know more? We’re happy to share our findings through a presentation or workshop. Get in touch at jenny@ournormal.org if you’d like to explore what these insights could mean for your organisation.

By AI assistants, we mean AI services and LLMs (large language models) that you can have a conversation with through text, voice or images – for example ChatGPT, Gemini, Claude, Copilot, Meta AI, Perplexity, Snapchat’s My AI or similar services.

Browse and download “AI assistants and neurodivergent youth”. If you use a screen reader, we recommend the text version file further down the page.

Links to the report