Addicted, attached, or just delegating? A scoping review on "problematic artificial intelligence use"

Area of Science:

• Artificial Intelligence
• Psychology
• Human-Computer Interaction

Background:

• Generative and conversational AI diffusion raises concerns about problematic use and dependence.
• A lack of a common framework hinders research despite a proliferation of studies.
• This scoping review addresses definitions, measurement, correlates, outcomes, and operationalization of problematic AI use.

Purpose of the Study:

• To map the definitions, measurement approaches, correlates, outcomes, and evidence of problematic AI use.
• To identify inconsistencies and heterogeneity in the conceptualization and operationalization of problematic AI use.
• To synthesize findings across three main strands: behavioral addiction, cognitive over-reliance, and psychological dependence.

Main Methods:

• Scoping review conducted following PRISMA-ScR guidelines.
• Searched Web of Science and Scopus databases using predefined strings for artificial intelligence and problematic use.
• Included 37 empirical peer-reviewed studies.

Main Results:

• Inconsistent terminology and heterogeneity in conceptualizing problematic AI use were found.
• Behavioral addiction is linked to depression, loneliness, social anxiety, and lower self-esteem, with younger age and male gender as risk factors.
• Cognitive over-reliance correlates with performance expectations and skill erosion, while psychological dependence is associated with loneliness and anthropomorphism.

Conclusions:

• The field of problematic AI use is fragmented, requiring clearer construct specification and AI-specific validated scales.
• Longitudinal and experimental designs are needed to establish causal mechanisms.
• Recommendations support safer system design and a better understanding of AI dependence.