Using Artificial Intelligence to Improve Empathetic Statements in Autistic Adolescents and Adults: A Randomized Clinical Trial

Challenges with social communication and social interaction are a defining characteristic of autism spectrum disorder (ASD). These challenges frequently interfere with making friendships, securing and maintaining employment, and can lead to co-occurring conditions. While face-to-face clinical interventions with trained professionals can be helpful in improving social conversation, they can be costly and are unavailable to many, particularly given the high prevalence of ASD and lack of professional training. The purpose of this study was to assess whether an AI program using a Large Language Model (LLM) would improve verbal empathetic responses during social conversation. Autistic adolescents and adults, 11-35 years of age, who were able to engage in conversation but demonstrated challenges with empathetic responses participated in this study. A randomized clinical trial design was used to assess the effects of the AI program (Noora) compared to a waitlist control group. Noora asks participants to respond to leading statements and provides feedback on their answers. In this study, participants were asked to respond to 10 statements per day 5 days per week for 4 weeks for an expected total of 200 trials. Pre- and post-intervention conversation samples were collected to assess generalization during natural conversation. Additionally pre- and post-intervention questionnaires regarding each participant's comfort during social conversation and participants' satisfaction with the AI program were collected. The results of this study demonstrated that empathetic responses could be greatly improved by using an AI program for a short period of time. Participants in the experimental group showed statistically significant improvements in empathetic responses, which generalized to social conversation, compared to the waitlist control group. Some participants in the experimental group reported improved confidence in targeted areas and most reported high levels of satisfaction with the program. These findings suggest that AI using LLMs can be used to improve empathetic responses, thereby providing a time- and cost-efficient support program for improving social conversation in autistic adolescents and adults.

Authors

Institutions

Publication Details

Journal
Journal of Autism and Developmental Disorders
Published
2025-02-15
DOI
https://doi.org/10.1007/s10803-025-06734-x
Citations
4
Primary Topic
Empathy and Medical Education
Type
article
Field-Weighted Citation Impact
5.88
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
article

Using Artificial Intelligence to Improve Empathetic Statements in Autistic Adolescents and Adults: A Randomized Clinical Trial

Monica S. Lam, Meng-Li Wang, Elizabeth L. Ponder, Nathan Chi et al.
4 citations
Journal of Autism and Developmental Disorders
Empathy and Medical Education
5.88
article

Using Artificial Intelligence to Improve Empathetic Statements in Autistic Adolescents and Adults: A Randomized Clinical Trial

Monica S. Lam, Meng-Li Wang, Elizabeth L. Ponder, Nathan Chi, Tzu‐Ying Lin, Brittany L. Koegel, Lynn Koegel, Shahryar Semnani, Tommy Bruzzese, Ankush Swarnakar
article en
4 citations

Abstract

Challenges with social communication and social interaction are a defining characteristic of autism spectrum disorder (ASD). These challenges frequently interfere with making friendships, securing and maintaining employment, and can lead to co-occurring conditions. While face-to-face clinical interventions with trained professionals can be helpful in improving social conversation, they can be costly and are unavailable to many, particularly given the high prevalence of ASD and lack of professional training. The purpose of this study was to assess whether an AI program using a Large Language Model (LLM) would improve verbal empathetic responses during social conversation. Autistic adolescents and adults, 11-35 years of age, who were able to engage in conversation but demonstrated challenges with empathetic responses participated in this study. A randomized clinical trial design was used to assess the effects of the AI program (Noora) compared to a waitlist control group. Noora asks participants to respond to leading statements and provides feedback on their answers. In this study, participants were asked to respond to 10 statements per day 5 days per week for 4 weeks for an expected total of 200 trials. Pre- and post-intervention conversation samples were collected to assess generalization during natural conversation. Additionally pre- and post-intervention questionnaires regarding each participant's comfort during social conversation and participants' satisfaction with the AI program were collected. The results of this study demonstrated that empathetic responses could be greatly improved by using an AI program for a short period of time. Participants in the experimental group showed statistically significant improvements in empathetic responses, which generalized to social conversation, compared to the waitlist control group. Some participants in the experimental group reported improved confidence in targeted areas and most reported high levels of satisfaction with the program. These findings suggest that AI using LLMs can be used to improve empathetic responses, thereby providing a time- and cost-efficient support program for improving social conversation in autistic adolescents and adults.

Journal of Autism and Developmental Disorders
Stanford Medicine (US), Stanford University (US)
Quality Education
Openalex Percentile: Top 5%
Empathy and Medical Education
5.88
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.