Agentic AI: Political Decisions Behind Technical Arrangements
A recent study published on arXiv (2608.06510) explores the political aspects of agentic AI, contending that the interests these systems prioritize are frequently determined by technical factors such as API selections, governance protocols, industry standards, and default settings. The authors perform a comparative analysis of four established areas—browser-based ad blockers, platform recommendation systems, financial robo-advisors, and email spam management—to demonstrate how similar types of agency have been influenced. They introduce the term 'depoliticization' from political theory to explain how these technical choices obscure their political implications. The research indicates that the potential for empowerment through agentic AI relies on more than just technology, as individual outcomes and collective challenges are shaped by these concealed political decisions. This study contributes to ongoing discussions regarding AI governance and user independence.
Key facts
- Paper on arXiv: 2608.06510
- Comparative case analysis of ad blockers, recommender systems, robo-advisors, and email spam governance
- Identifies depoliticization as a key concept
- Technical arrangements include API choices, protocol governance, industry standards, and default configurations
- Argues that agentic AI's impact on user empowerment is an open question
- Decisions about whose interests agents serve are political, not just technical
- The paper is a cross-type announcement on arXiv
- Focus on digital agency and user empowerment
Entities
Institutions
- arXiv