Artificial Intelligence · 28.08.2026, 07:02 UTC
Toward a New Science of AI as Cognitive Infrastructure
| Schweregrad | info |
|---|---|
| Kategorie | Artificial Intelligence |
| Quelle | arXiv cs.AI ↗ |
| Veröffentlicht | 28.08.2026 UTC |
Sicherheitsmeldung mit Schweregrad noch nicht bewertet. Technische Details im Tab „Originaltext“; empfohlene Schritte in der Checkliste.
arXiv:2507.22893v3 Announce Type: replace-cross Abstract: Contemporary human-AI interaction research overlooks how AI systems fundamentally reshape human cognition pre-consciously, a critical blind spot for understanding distributed cognition. This paper introduces "Cognitive Infrastructure Studies" (CIS) as a new interdisciplinary domain to reconceptualize AI as "cognitive infrastructures": foundational, often invisible systems conditioning what is knowable and actionable in digital societies. These semantic infrastructures transport meaning, operate through anticipatory personalization, and exhibit adaptive invisibility, making their influence difficult to detect. Critically, they automate "relevance judgment," shifting the "locus of epistemic agency" to non-human systems. Through narrative scenarios spanning individual (cognitive dependency), collective (democratic deliberation), and societal (governance) scales, we describe how cognitive infrastructures reshape human cognition, public reasoning, and social epistemologies. CIS aims to address how AI preprocessing reshapes distributed cognition across individual, collective, and cultural scales, requiring unprecedented integration of diverse disciplinary methods. The framework also addresses critical gaps across disciplines: cognitive science lacks population-scale preprocessing analysis capabilities, digital sociology cannot access individual cognitive mechanisms, and computational approaches miss cultural transmission dynamics. To achieve this goal CIS also provides methodological innovations for studying invisible …
Maßnahmen
⬇ Als MarkdownVerwandte Beiträge
- info Scaling Model-Generated Distillation Data Can Make Latent Teacher Traits More Recoverable
- info A Layer Importance Metric for Quantization Accounting for the Speed-Quality Trade-off in Autoregressive Models
- info When Privacy Hurts Mergeability: Geometry-Aware Model Merging under Differential Privacy
- info GRAS: Guided Reduced-Variance Proposals and Adaptive Selection for Training-Free Reward Alignment in Discrete Diffusion