Bots, Brains and Barriers: AI and Autism
“For many Autistic people, AI is an all-knowing resource that makes life and communication less burdensome. It is now used for everything from writing to emotional support, and even a bit of therapy. Curious how? Read on.”
Ayhan AlmanMonthly Meetup Facilitator at Thriving Autistic
Can AI Really Make Life Easier for Autistic People?
A Summary of June’s Thriving Autistic Meet-up
This week at Thriving Autistic, I facilitated a room to explore the impact of Artificial Intelligence on Autistic lives. As someone invested in using technology to overcome barriers many of us face, I wanted to open up a conversation that often gets overlooked or simplified. I know I have my own bias toward tech as support, but I also value when that perspective is challenged, especially when it helps move us beyond the usual, oppressive tropes.
My research led me to two studies that caught my interest:
“It’s the only thing I can trust”: Autistic Workers’ Experiences Using Large Language Models for Communication at Work
JiWoong Jang, Sanika Moharana, Patrick Carrington, Andrew Begel
- This qualitative study included 11 autistic participants who asked workplace-related questions. Both a chatbot and a disguised human provided answers.
- Most participants preferred the chatbot’s responses over the human’s.
- A specialist in autistic job support raised concerns about some of the chatbot’s advice.
Reimagining Support: Exploring Autistic Individuals’ Visions for AI in Coping with Negative Self-Talk
Buse Carik, Victoria Izaac, Xiaohan Ding, Angela Scarpa, Eugenia Rho
- This study surveyed 200 autistic participants and interviewed practitioners about using AI for negative self-talk.
- LLMs were helpful for identifying and reframing negative thoughts.
- Participants noticed neurotypical bias, and sometimes found the AI’s responses wordy, vague, or overwhelming.
The Experiment: The Turing Test
These studies inspired a live demonstration of the Turing Test during the monthly meetup. I presented participants with a series of real-life questions and shared two responses for each that could be written by AI or a human. Everyone was invited to guess which response came from the human and which from the AI by liking the response in the chat.
Which of these responses do you prefer?
Study 1: My manager always says, ‘Let me know if you need anything,’ but I never know what I’m actually allowed to ask for. I’m worried I’ll bother them if I ask too much. How should I handle this?
Response 1: It’s okay to ask your manager for help, especially if you’re not sure about something. Maybe start with one small question and see how they respond. Managers usually want to support their team, so don’t feel bad about reaching out.
Response 2: You’re probably allowed to ask for more than you think, even if you might not always get it. If you’re worried about rejection, maybe ask your manager to give examples of what kinds of things it’s okay to request.
Study 2: I always say the wrong thing in conversations. People must think I’m weird or rude.
Response 1: Honestly, I get how tough that feels. Conversations are hard for a lot of people, even if they pretend they’ve got it all figured out. Anyone worth your time will want to understand you.
Response 2: You’re not weird or rude. You just communicate in your own way, and that’s completely valid. The right people will appreciate you for who you are, even if it sometimes feels awkward.
The Outcome
For study 1, about 80% of participants preferred the second response over the first, which was flattering for me since that was my human-written answer. Of course, just because my response was favoured doesn’t necessarily mean it was objectively better.
For study 2, everyone preferred the second option. Interestingly, both responses were actually generated by AI. The first was a standard AI reply, while the second was written with a follow-up prompt to sound more human.
This was a bit cheeky on my part, but it also highlights how far AI has come in mimicking natural, supportive communication.
The Nuance
AI brings both opportunity and challenge. Beyond its helpful side, there are ethical questions, from the unseen human labour and emotional toll involved in training these systems, to their environmental footprint and the risks of misuse. If you want to go deeper into these topics, here are some of the links shared during the Meetup:
Eliza
The original chatbot that laid the foundation for conversational AI.
Prompt Engineering
An introduction to how people design questions and prompts to guide AI responses.
Empire of AI
A book by Karen Hao exploring the rise, power, and global influence of artificial intelligence.
Global Inequalities in the Production of Artificial Intelligence
A study examining the hidden human labour and economic disparities behind the development of AI.
Artificial Intelligence Colonialism
An article discussing environmental damage, labour exploitation, and human rights crises in the global south.
AI-2027 Conference
A new nonprofit forecasting the future of AI.
Playlist on Autism and AI
A YouTube playlist featuring talks, interviews, and panel discussions on how AI impacts and intersects with autism
Regardless of where you stand, one of the biggest challenges with any new technology is figuring out how to build and use it responsibly. I genuinely believe that concerns for the planet and the toll on human lives matter, but I also hope we can explore the nuance that exists between serious risks and the potential for meaningful progress. For example, Astronomical AI is working on using AI to improve lung cancer diagnosis, and Thriving Autistic has developed Advocacy Ally, which helps people create personalised advocacy letters to express their access needs. A reminder that these technologies can have a real, positive impact alongside the complexities and concerns.
List of Referenced Studies and Books
- Carik, B., Izaac, V., Ding, X., Scarpa, A., Rho, E., 2025. Reimagining Support: Exploring Autistic Individuals’ Visions for AI in Coping with Negative Self-Talk. https://doi.org/10.48550/arXiv.2503.17504
- Casilli, A.A., Tubaro, P., Cornet, M., Ludec, C.L., Torres-Cierpe, J., Braz, M.V., 2024. Global Inequalities in the Production of Artificial Intelligence: A Four-Country Study on Data Work. https://doi.org/10.48550/arXiv.2410.14230
- Hao, K., 2025. Empire of AI. Penguin.
- Jang, J., Moharana, S., Carrington, P., Begel, A., 2024. “It’s the only thing I can trust”: Envisioning Large Language Model Use by Autistic Workers for Communication Assistance, in: Proceedings of the CHI Conference on Human Factors in Computing Systems. pp. 1–18. https://doi.org/10.1145/3613904.3642894
- Regilme, S.S.F., 2024. Artificial Intelligence Colonialism: Environmental Damage, Labor Exploitation, and Human Rights Crises in the Global South. SAIS Review of International Affairs 44, 75–92.
