A new study shows that people with autism are using AI tools like ChatGPT to deal with workplace issues. Researchers at Carnegie Mellon University found that while AI can help, it sometimes gives questionable advice. This has sparked debate in the autism community about whether chatbots are useful.
Andrew Begel from CMU’s School of Computer Science shared, “People with autism are already asking ChatGPT questions like, ‘How do I make friends at work?’ ChatGPT can be both good and bad at answering these.”
Begel leads the VariAbility Lab, which focuses on making workplaces better for people with autism and other neurodivergent conditions. Many people with autism face challenges finding and keeping jobs, partly because workplaces don’t always have the right resources to help with social and communication issues.
The study involved 11 autistic participants who were given advice from two sources: a chatbot like ChatGPT and a human pretending to be a chatbot. Surprisingly, most participants preferred the chatbot. Begel thinks this is because the chatbot gave quick and simple answers, often in bullet points. In contrast, the human tried to dig deeper by asking more questions, which many participants didn’t want.
One participant said they trusted the chatbot because not all workplaces are inclusive.
However, an expert who reviewed the AI advice found that some suggestions weren’t helpful. For example, when asked how to make friends, the chatbot suggested just starting a conversation, which can be uncomfortable for many people with autism.
Begel believes that with better training, AI could give more suitable advice. But he also acknowledges that opinions in the autism community are divided. Some see chatbots as helpful, while others feel they push autistic people to adapt in ways that don’t align with how their brains work.
“There’s a big debate about whose needs are prioritized
Source: tomorrowsworldtoday
