arXiv:cs.AI· Ayoub Merimi, Souhail Bakkali, Abdeljalil Sakat·· 17 小时前
LLM 决策新鲜度研究:Fresh at Decision, Stale at Action 提出时间戳决策有效性理论
Fresh at Decision, Stale at Action: Routing New Information to LLM Decisions Already Made
AI 导读
研究者提出时间戳决策有效性理论与决策账本(decision ledger),用六个时间戳定义一项决策,并设新鲜度上限界定行动时最新证据的保证范围。在五个 LLM(从 8B 到前沿规模)上验证四项预测:仅凭新鲜度即达到上限预测的准确率,模型在关键消息到达后会据其行动。剩余错误集中在变化何时生效的判断上。
正文
Abstract:Freshness shapes an LLM decision at the moment it is made; the action it commits to is latent, carried out later, in a changing world. We therefore introduce a timestamp theory of decision validity and a decision ledger that realises it. Six timestamps define a decision; a freshness ceiling bounds what up-to-date evidence guarantees at action time, placing the remaining risk in information published afterwards. Accuracy then factorises into two capabilities: routing, which delivers that information to affected decisions before their revision deadlines, and temporal judgment, which reads it for the time of action. The ledger is persistent state holding each decision's times, evidence and background knowledge, indexed by watch conditions written at decision time. Controlled decision streams with exact labels confirm four predictions across five LLMs, from 8B to frontier scale: freshness alone yields exactly the accuracy the ceiling predicts; models act on decisive news once it arrives, so accuracy follows routing; the ledger's index sets what routing reaches, and watch conditions reach indirect dependencies at low exposure; and the remaining errors concern when a change takes effect. Freshness makes a decision right when it is made; routing and temporal judgment keep it right when it is carried out.
| Comments: | A timestamp theory of LLM decision validity with a decision ledger |
| Subjects: | Artificial Intelligence (cs.AI); Computer Science and Game Theory (cs.GT); Information Retrieval (cs.IR); Multiagent Systems (cs.MA) |
| ACM classes: | I.2.7; I.2.11; I.2.3; I.2.4; H.3.3 |
| Cite as: | arXiv:2610.18126 [cs.AI] |
| (or arXiv:2610.18126v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2610.18126 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Ayoub Merimi [view email]
[v1]
Tue, 13 Oct 2026 10:37:12 UTC (73 KB)
来源:arXiv:cs.AI · arxiv.org