arXiv：cs.AI· Sean Koon·· 4 小时前AI 评分26
以人为中心的混合推理系统框架：优先增强人类智能
Hybrid Reasoning Systems That Prioritize and Enhance Human Intelligence
AI 导读
一篇 arXiv 论文提出以人为中心的混合推理系统框架，通过整合人类推理增强策略、偏好"结论前介入"的 AI 设计方法，以及将推理视为多种可独立支持模式的处理方式，实现从细粒度数据分析到高层反思的广谱推理增强。
正文
Abstract:In a world of accelerating change, there is a need for wise and adaptive human reasoning. Integrating AI capabilities with human guidance offers promise, though human reasoning itself is often hasty, shortsighted, and error-prone, and no clear framework exists for combining human reasoning strategies with AI across diverse tasks. This article proposes a framework for human-centered hybrid reasoning systems that engage and enhance human reasoning abilities ranging from granular data analysis to high-level reflection and wisdom. The framework was developed through a conceptual synthesis combining: (1) established strategies for enhancing human reasoning, (2) AI design approaches that favor pre-conclusive engagement over the generation of conclusions, and (3) the treatment of reasoning as a collection of distinct, individually supportable modes. This synthesis produced a distinctive framework for broad-spectrum reasoning enhancement from which a typology of reasoning modes, a system architecture, and a high-level research agenda were derived.
| Comments: | 20 pages; 6 figures; 2 tables. This paper underwent significant extension and revision from prior archived version |
| Subjects: | Human-Computer Interaction (cs.HC); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2504.13477 [cs.HC] |
| (or arXiv:2504.13477v3 [cs.HC] for this version) | |
| https://doi.org/10.48550/arXiv.2504.13477 arXiv-issued DOI via DataCite |
|
| Journal reference: | International Journal of Human Computer Interaction, Page 1 to 19 (2026) |
| Related DOI: | https://doi.org/10.1080/10447318.2026.2735159
DOI(s) linking to related resources |
Submission history
From: Sean Koon [view email]
[v1]
Fri, 18 Apr 2025 05:38:21 UTC (722 KB)
[v2]
Wed, 14 Jan 2026 05:50:29 UTC (941 KB)
[v3]
Fri, 2 Oct 2026 17:21:22 UTC (704 KB)
来源：arXiv：cs.AI · arxiv.org