Modeling public trust in AI cognitive capabilities using statistical and machine learning approaches

Area of Science:

• Human-computer interaction
• Artificial intelligence ethics
• Cognitive science

Background:

• Artificial intelligence (AI) systems are increasingly performing cognitive tasks.
• Assessing public trust in AI's cognitive abilities is crucial for adoption.
• Understanding factors influencing trust is key for developing user-aligned AI.

Purpose of the Study:

• To investigate the impact of age, gender, and AI familiarity on public confidence in AI's cognitive functions.
• To identify key predictors of trust in AI decision-making, judgment, and memory recall.
• To explore user preferences for AI versus human decision-making in different scenarios.

Main Methods:

• Survey of 400 participants on AI trust and familiarity.
• Statistical analysis of demographic and experiential factors.
• Random Forest classification to predict trust levels.

Main Results:

• AI familiarity was the strongest predictor of confidence in AI.
• Age and gender also significantly influenced trust levels.
• Participants trusted AI more for factual recall than complex judgments or high-stakes decisions.

Conclusions:

• Familiarity and age are the most influential factors in shaping trust in AI's cognitive capabilities.
• Public trust in AI is nuanced, varying by task type and perceived risk.
• Findings offer practical insights for designing trustworthy and user-centered AI systems.