On the Influence of Artificial Intelligence on Human Problem-Solving: Empirical Insights for the Second Wave within a Multinational Longitudinal Pilot Study

The swift integration of artificial intelligence and large language models into daily work is changing how humans solve problems. Following Wave 1, which identified context-dependent ethics and fragile trust, this paper shares intermediate results from Wave 2 of a multinational longitudinal study. A diverse sample of 36 people completed a comprehensive survey and four graded problem vignettes. This captured their demographics, AI proficiency, usage patterns, ethical perceptions, verification capabilities, and performance. Findings point to a consolidating hybrid problem-solving culture. This culture uses a structured process where after an initial human framing, AI is strategically embedded for tasks including research at 72.2 percent, checking correctness at 69.4 percent, ideation at 61.1 percent, and formulation at 61.1 percent. Consultation with AI increases as tasks become more complex, rising from 11.1 percent on a simple task to 48.1 percent on a complex probability task and 44.0 percent on an optimization task. However, actual performance declines with this complexity, as accuracy drops from 90.3 percent to 46.7 percent. Two systematic gaps also widen with complexity. One is a belief-performance gap where perceived correctness exceeds actual accuracy by 34.6 percentage points on the most complex tasks. The other is a proof-belief gap where verification capability lags behind confidence by 17.7 percentage points. Together with Wave 1, these results indicate a shift toward distributed human-AI cognition that is accompanied by verification deficits. We outline planned Wave 3 interventions including verification scaffolds, trust-calibration training, and task-specific proof requirements. These are designed to strengthen critical AI literacy and ensure reliable outcomes in complex problem-solving environments.