Mitigating Societal Cognitive Overload in the Age of AI: Challenges and Directions

Cognitive overload is a foundational AI safety issue: The paper argues that societal cognitive overload—caused by information proliferation, algorithmic manipulation, and automation anxieties—is not just a quality-of-life problem but a critical prerequisite for managing both current AI harms and potential future risks from advanced AI systems.

AI actively worsens information overwhelm through multiple mechanisms: Beyond simply creating more information, AI exacerbates cognitive overload through algorithmic manipulation, deregulation of content, erosion of shared meaning, and anxiety about automation, making it harder for people to make informed decisions about technology itself.

Cognitive overload undermines human-AI alignment: When people are cognitively overwhelmed, they cannot meaningfully participate in decisions about AI governance, understand AI risks, or hold institutions accountable—creating a vicious cycle where the very conditions needed for responsible AI development are being eroded.

Institutional and policy changes are needed: The paper calls for institutional adaptations and policy considerations designed to build "overload-resilience," though it frames these as directions for exploration rather than ready-made solutions, suggesting this is an emerging area requiring collaborative problem-solving.

This reframes the AI safety debate: By centering cognitive overload, the paper bridges near-term harms (misinformation, decision fatigue, anxiety) with long-term risks, suggesting that addressing how AI overwhelms human cognition today is essential preparation for managing more advanced AI systems tomorrow.