Overview
This research presents a lifecycle-oriented synthesis of sociotechnical risks impacting human-AI collaboration. The study identifies six recurring cross-domain risk clusters and proposes a conceptual interaction model to illustrate their emergence from sociotechnical drivers, cascading pathways, and impact on team performance and human well-being. The analysis suggests that collaboration failures often result from interconnected sociotechnical dynamics rather than isolated technical deficiencies.
Research Context
AI systems are increasingly integrated into consequential domains, including healthcare, journalism, education, scientific research, organizational decision-making, and defense. This expanded integration has made effective human-AI collaboration a critical challenge. Existing research often studies sociotechnical risks that undermine collaboration in isolation, which may obscure recurring failure mechanisms present across diverse domains. The current work aims to synthesize this fragmented literature into a unified framework.
Approach
The research adopted a lifecycle-oriented approach to analyze human-AI collaboration risks. This involved spanning four distinct stages of collaboration: task allocation, interaction, feedback, and adoption. Evidence was drawn from diverse application domains to identify recurring patterns. The methodology further involved proposing a conceptual interaction model to visualize how identified risks emerge from sociotechnical drivers, interact through cascading pathways, and ultimately influence team performance and human well-being.
Findings
The research identified six recurring cross-domain risk clusters that impact human-AI collaboration:
- Trust Miscalibration: This cluster refers to issues where human trust in AI is either too high or too low, leading to inappropriate reliance or underutilization.
- Cognitive Burden: This involves increased mental effort or cognitive load on human collaborators due to interactions with AI systems.
- Accountability Gap: This refers to situations where responsibility for errors or failures within human-AI collaborative systems is unclear or unassigned.
- Capability Erosion: This cluster describes the potential for human skills or capabilities to diminish over time due to over-reliance on AI systems.
- Goal Misalignment: This occurs when the objectives or incentives of the human and AI components of a system are not adequately aligned.
- AI Anxiety and Technostress: This cluster encompasses psychological stress, discomfort, or negative emotional responses experienced by humans when interacting with AI technologies.
These risk clusters were found to emerge across the four collaboration lifecycle stages: task allocation, interaction, feedback, and adoption. The study's conceptual interaction model illustrates that these risks originate from sociotechnical drivers and can interact through cascading pathways. This interaction ultimately affects team performance and human well-being. The analysis indicated that many collaboration failures stem from interconnected sociotechnical dynamics rather than isolated technical deficiencies.
Why This Matters
This work provides a foundation for future empirical research, lifecycle-oriented governance, and the design of more resilient, trustworthy, and human-centered human-AI collaboration systems. By synthesizing fragmented literature, the research offers a unified framework that helps explain why piecemeal interventions often lead to unintended consequences in human-AI collaboration.