Self-guided mental health interventions offer promise but are often cognitively demanding and emotionally triggering. This paper studies how human-language model interaction can support cognitive restructuring, an evidence-based therapeutic technique. A randomized field study on a large mental health website (15,531 participants) evaluated a system using language models to support cognitive restructuring. Findings show positive impacts on emotional intensity for 67% of participants and help overcoming negative thoughts for 65%. While adolescents showed worse outcomes, tailored interventions simplifying language model generations improved overall effectiveness and equity.
Publisher
CHI '24: ACM CHI Conference on Human Factors in Computing Systems
Published On
May 11, 2024
Authors
Ashish Sharma, Kevin Rushton, Theresa Nguyen, Inna Wanyin Lin, Tim Althoff
