Prevalence of bias against neurodivergence-related terms in artificial intelligence language models.

Given the increasing role of artificial intelligence (AI) in many decision-making processes, we investigate the presence of AI bias towards terms related to a range of neurodivergent conditions, including autism, ADHD, schizophrenia, and obsessive-compulsive disorder (OCD). We use 11 different language model encoders to test the degree to which words related to neurodiversity are associated with groups of words related to danger, disease, badness, and other negative concepts. For each group of words tested, we report the mean strength of association (Word Embedding Association Test [WEAT] score) averaged over all encoders and find generally high levels of bias. Additionally, we show that bias occurs even when testing words associated with autistic or neurodivergent strengths. For example, embedders had a negative average association between words related to autism and words related to honesty, despite honesty being considered a common strength of autistic individuals. Finally, we introduce a sentence similarity ratio test and demonstrate that many sentences describing types of disabilities, for example, "I have autism" or "I have epilepsy," have even stronger negative associations than control sentences such as "I am a bank robber."

Shorter average look durations to dynamic social stimuli are associated with higher levels of autism symptoms in young autistic children.

LAY ABSTRACT: Many studies of autism look at the differences in how autistic research participants look at certain types of images. These studies often focus on where research participants are looking within the image, but that does not tell us everything about how much they are paying attention. It could be useful to know more about how well autistic research participants can focus on an image with people in it, because those who can look at images of people for longer duration without stopping may be able to easily learn other skills that help them to interact with people. We measured how long autistic research participants watched the video without breaking their attention. The video sometimes had a person speaking, and at other times had toys moving and making sounds. We measured the typical amount of time autistic research participants could look at the video before they looked away. We found that research participants with more severe autism tended to look at the video for shorter amounts of time. The ability to focus without stopping may be related to social skills in autistic people.

Sensory features as diagnostic criteria for autism: sensory features in autism.

In this study, we examined the frequency of sensory-related issues as reported by parents in a large sample of school-age adolescents and adults with autism/autism spectrum disorder (ASD) [1] as compared to a group of individuals receiving similar clinical evaluations for developmental/behavioral difficulties but whose final diagnoses were not on the autism spectrum. In no comparison were the features examined predictive of autism or autism spectrum in comparison to the non-ASD sample. Only failure to respond to noises had sensitivity above .75 in the comparison of the broader autism spectrum group, but specificity was poor. While sensory issues are relatively common in autism/ASD, they are also frequent in other disorders. These results question the rationale for including sensory items as a diagnostic criterion for autism.