AI Tools Like ChatGPT Offer Mixed Support for Autistic Workers, Prompting Debate on Approaches and Inclusivity
A recent study by Carnegie Mellon University highlights the increasing use of AI tools like ChatGPT by individuals with autism for workplace advice, revealing a preference for clear, concise responses over interactive counseling. However, the findings also raise concerns about the quality of advice provided and whether such tools align with the needs and preferences of the autism community. The study stresses the importance of involving autistic individuals in developing assistive technologies to ensure they address actual needs and preferences.
Key Points:
- Study Context:
- Conducted by Carnegie Mellon University’s School of Computer Science.
- Led by Andrew Begel, associate professor, focusing on workplace inclusivity for neurodivergent individuals.
- Participants:
- 11 individuals with autism recruited to test AI tools versus human counselors for workplace advice.
- Preferences and Findings:
- Participants preferred AI chatbot responses for their directness and simplicity.
- Chatbot responses were typically bullet-pointed and straightforward, which was favored over the interactive approach of human counselors.
- Human counselors tended to ask exploratory questions, which participants found less appealing.
- Concerns:
- Some advice from the chatbot was not practical for autistic individuals (e.g., advice to initiate conversations by approaching people directly).
- Evaluation by a professional indicated that some chatbot advice could be unhelpful or inappropriate.
- Community Perspectives:
- There is a split opinion within the autism community on the use of AI tools.
- Concerns about whether these tools enforce neurotypical standards on autistic individuals.
- Research Implications:
- Suggests the potential for AI tools tailored specifically to the needs of autistic users.
- Emphasizes the necessity of involving autistic individuals in the development process to ensure the relevance and usefulness of assistive technologies.
- Related Study:
- Analysis of 142 papers on human-robot interaction found 90% did not include autistic perspectives.
- Resulted in assistive technologies that may not meet the actual needs of the autism community.
- Future Directions:
- Ongoing projects include AI analysis of conversations to identify and address communication gaps.
- Development of tools to assist in real-time conversation management (e.g., asking for repetition or clarification).
- Creation of an Autism Advisory Board to guide research and technology development.
“What we found is there are people with autism who are already using ChatGPT to ask questions that we think ChatGPT is partly well-suited and partly poorly suited for.”
— Andrew Begel, Associate Professor in the Software and Societal Systems Department and the Human-Computer Interaction Institute, Carnegie Mellon University
More on Autism Spectrum Disorders (ASD)