Assessing AI-Supported Problem-Solving in Young Adults Reporting Previous Learning-Disorder Diagnoses: Adaptation and Validation of the PS-AIRES and Preliminary Evaluation of a Claude-Based Training Program

Abstract

AI-supported problem-solving environments increasingly require users to formulate effective prompts, iteratively refine interactions, critically evaluate outputs, and integrate AI assistance into coherent, self-directed solutions. Using a mixed-methods design, this two-phase study involved 672 participants across developmental, validation, and intervention stages. Study 1 included 12 participants in qualitative and cognitive interviewing, 40 independent participants in pilot testing, and 500 participants in the main validation sample. The study developed and psychometrically validated the Problem-Solving Skills in AI-Rich Environments Scale (PS-AIRES) and conducted a preliminary evaluation of a brief Claude-based training program. Exploratory and confirmatory factor analyses supported a three-factor model comprising Strategic Task Framing and AI Instruction Design, Iterative Prompting and Response Refinement, and Verification, Critical Evaluation, and Integration. The model demonstrated satisfactory fit, high internal consistency, and acceptable temporal stability (ICC = 0.713, 95% CI [0.479, 0.852]). Measurement invariance across gender supported structural stability. A subsequent randomized pretest–posttest study involving an independent sample of 120 young adults reporting a previous specialist diagnosis of a learning disorder preliminarily evaluated the brief Claude-based training program. The findings provide initial evidence for the reliability and validity of PS-AIRES and suggest the short-term usefulness of structured, dyslexia-sensitive AI literacy training for enhancing problem-solving in AI-rich environments.

The authors' abstract, as published at the source. International Journal of Human-Computer Interaction, 2026 · DOI ↗