ChatGPT Helps Students Feign ADHD: An Analogue Study on AI-Assisted Coaching

Fuermaier, Anselm B. M.; Niesten, Isabella J. M. · 2025 · Crossref

DOI: 10.1007/s12207-025-09538-7

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This preregistered analogue study investigates whether artificial intelligence (AI) chatbots, specifically ChatGPT, can effectively coach individuals to feign attention-deficit/hyperactivity disorder (ADHD) during clinical neuropsychological assessments. The research addresses a growing concern that publicly accessible AI tools may compromise the validity of ADHD diagnoses by providing examinees with sophisticated strategies to mimic symptoms while evading detection. The authors note that adult ADHD assessments rely heavily on self-reports and cognitive tests, which are susceptible to distortion. While previous research has examined the impact of online information and traditional coaching, this study specifically evaluates the efficacy of AI-generated instructions in helping non-clinical individuals simulate ADHD convincingly. The study involved 110 university students randomly assigned to one of three groups: a control group instructed to perform normally, a symptom-coached simulation group provided with DSM-5 diagnostic criteria, and an AI-coached simulation group provided with a concise information sheet generated by ChatGPT. The AI coaching material was developed by querying ChatGPT with 87 questions formulated by students and researchers regarding how to feign ADHD and avoid detection. All participants underwent a clinical assessment battery including the Conners’ Adult ADHD Rating Scale (CAARS), the Weiss Functional Impairment Rating Scale (WFIRS), the Perceptual and Attention Functions–Selective Attention test (WAFS), and the Reliable Digit Span (RDS). The assessment included embedded symptom validity tests (SVTs) and performance validity tests (PVTs) to detect feigning. Results indicated that both simulation groups reported significantly higher ADHD symptoms and impairments and performed worse on cognitive tests compared to the control group. However, the AI-coached group demonstrated more nuanced and credible feigning strategies than the symptom-coached group. Specifically, the AI-coached participants moderated their symptom overreporting and cognitive underperformance, resulting in lower scores on validity indices such as the CAARS Infrequency Index and ADHD Credibility Index. This moderation led to significantly lower detection sensitivity for the AI-coached group compared to the symptom-coached group. The AI-generated instructions explicitly advised against extreme exaggeration, instead promoting consistency and subtlety to appear more convincing. The findings suggest that publicly accessible AI tools pose a significant threat to the validity of clinical ADHD assessments by providing effective, strategy-based coaching that helps individuals evade detection. The study concludes that AI can combine symptom knowledge with test-specific strategies to produce credible feigned presentations. Consequently, the authors recommend that clinicians and researchers exercise caution when sharing assessment materials, example items, and scoring methodologies, as such information can be leveraged by AI to generate effective feigning guides. This highlights the need for updated approaches to validity testing and test security in the era of generative AI.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-25; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• illicit drugs