What is it about?
Agentic AI systems can increasingly act on our behalf, making decisions and carrying out multi-step tasks with limited human intervention. But as delegating tasks to AI becomes socially normal, people may also change how they judge whether that delegation is ethically acceptable. This study investigates whether growing social acceptance of AI delegation can weaken the role of personal moral judgment. Using survey data from 280 European university students and structural equation modeling, we examine how moral disengagement, perceived social norms, and shame proneness influence people’s intention to delegate academic tasks to AI. The results reveal a “normative substitution” effect: social norms both increase willingness to delegate to AI and reduce the relative influence of internal moral regulation. However, individuals with stronger shame proneness appear more resistant to this social pressure
Featured Image
Why is it important?
As agentic AI becomes embedded in everyday work, education, and decision-making, responsible AI cannot focus only on what the technology is capable of doing. It must also consider how widespread AI use may gradually change human moral responsibility. Our findings suggest that when AI delegation becomes socially accepted, people may increasingly use “what others consider acceptable” as a substitute for their own internal moral evaluation. This creates a subtle governance risk: ethically questionable delegation may become normalized even when individuals would previously have considered it inappropriate. The study therefore highlights the need for AI systems, institutions, and governance frameworks to preserve meaningful human accountability rather than allowing social normalization to become an automatic justification for delegation.
Perspectives
The ethical challenge of agentic AI is not only whether an AI system makes good or bad decisions. It is also whether repeated delegation changes the way humans themselves make moral judgments. Our results suggest that increasing social acceptance of AI can create a kind of “social autopilot,” where perceived norms progressively replace internal moral gatekeeping. This does not mean that people lose morality, but rather that the social environment can become a stronger reference point for deciding whether delegation is acceptable. For responsible human–AI collaboration, designers and policymakers should therefore pay attention not only to AI autonomy, transparency, and technical safeguards, but also to the social processes through which delegation becomes normalized.
Prof. Afshin Ashofteh
Universidade Nova de Lisboa
Read the Original
This page is a summary of: From Moral Gatekeeping to Social Autopilot: Revealing the Normative Substitution Paradox in AI Delegation, July 2026, Institute of Electrical & Electronics Engineers (IEEE),
DOI: 10.1109/compsac69091.2026.00162.
You can read the full text:
Contributors
The following have contributed to this page







