Malingering depression: a comparative study of human and GPT-3.5 performance

Key Points

• The assessment of malingering depression revealed significant performance differences between human evaluators and GPT-3.5 models, highlighting AI's capabilities.
• During evaluations, humans accurately identified 75% of cases, whereas GPT-3.5 demonstrated a 65% detection rate in simulated scenarios.
• A comparative evaluation was employed, testing both human and AI performance against standardized depression indicators.
• Findings suggest that while GPT-3.5 performs well, further validation and improvements are needed for reliable mental health assessments.