1
A Theory of Post-hoc Debate Judgement
从理论层面拆解事后辩论评判机制，给AI对齐与评估研究提供全新视角。
arXiv:2608.19002v1 Announce Type: new Abstract: Debates have recently emerged as a useful methodology for agentic AI to improve performance as well as…
2
Asymptotic Risk Calibration for Selective Question Answering
不看全文也能“有把握放弃回答”？这项研究为选择性问答系统提供了风险校准的理论保障。
arXiv:2608.12008v1 Announce Type: new Abstract: Large language models (LLMs) may generate fluent but incorrect answers, making uncertainty quantificat…
3
Power and Limitations of Aggregation in Compound AI Systems
复合AI系统聚合能力的边界在哪？这篇论文给出理论答案，值得关注AI架构的开发者细读。
arXiv:2602.21556v2 Announce Type: replace Abstract: When designing compound AI systems, a common approach is to query multiple copies of the same mode…
4
Show HN: Signal for LLM – The Modulator Architecture (Theory Complete)
把信号调制引入大模型，理论闭环完成，探索信息传递新范式。
Article URL: https://divinecanon.github.io/signalengine-EN/ Comments URL: https://news.ycombinator.com/item?id=48819823 Points: 2 # Comments: 0
5
VeriBound: PAC-Bayesian Generalization Bounds for Process Reward Models Trained with Formal Verification Tools
将形式化验证与PAC-Bayesian理论结合，为过程奖励模型提供可证实的泛化边界，是AI安全领域的硬核进展。
arXiv:2606.20740v1 Announce Type: cross Abstract: Process Reward Models (PRMs) provide step-level verification for Large Language Model (LLM) reasonin…
6
Emergent Language as an Approach to Conscious AI
探讨涌现语言如何作为实现有意识AI的新路径，融合语言学与认知科学前沿，挑战传统AI意识理论。
arXiv:2606.06380v1 Announce Type: cross Abstract: The question of whether artificial systems can be conscious remains open, in part because existing a…