Artificial Intelligence
给 Agent 加了记忆，它反而变笨了
记忆几乎总是被当作纯增益功能加上去的，但 2026 年的多份公开测量显示：加了记忆的 Agent 在不少任务上比不加更差。本文拆解稀释、误差累积、陈旧状态三种退化机制，对比 compaction、结构化笔记、子 Agent 隔离三种手段的取舍，并给出写入侧的过滤与遗忘策略和三组可自测的对照实验。
Topic
Artificial Intelligence
Technical progress of models, algorithms and intelligent systems and their industrial impact.
Featured article
Latest articles
Artificial Intelligence
Before You Delegate to AI, Decide What Must Stay Human
Starting with an agentic system’s ability to act in the external world, this article explains how consequences, permissions, data sensitivity and recoverability define delegation boundaries—and how approval, responsibility red lines, least privilege, traceable actions and rehearsed recovery constrain AI action.
Artificial Intelligence
How AI Actually Improves Productivity at Work: By Cutting Waiting, Switching and Rework
Drawing on research in writing, customer support, workplace collaboration and software development, this article explains when AI speeds direct tasks, when it shifts time into verification and rework, and how teams should measure net efficiency.
Artificial Intelligence
When AI Gives the Answer, Who Will Still Visit the Original Website?
Using Pew browsing data and generative-search research within their stated limits, this article separates answer satisfaction, source opening, verification and task completion, then proposes a falsifiable framework for page responsibility and measurement.
Artificial Intelligence
When AI Starts Working for You, the Real Change Is Not Efficiency
This article distinguishes task automation from workflow delegation, examines how continuous AI execution changes judgement, permissions, evidence, accountability and skill development, and proposes a practical five-step governance framework.
Artificial Intelligence
The EU AI Act Enters Its Implementation Phase: An Enterprise Compliance Guide
As the EU AI Act enters its implementation phase, compliance is becoming a basic capability for entering the European market and operating AI there over time. This guide follows four arguments: responsibility boundaries, risk-based controls, evidence across the supply chain, and governance embedded in everyday operations.
Artificial Intelligence
From 128K to 1M: What Long-Context Workflows Save and What They Add
Put long context back into concrete work such as contract review, cross-version comparison, and technical decisions. This article explains what 128K and 1M actually solve, and why retrieval, citations, and version control still matter.
Artificial Intelligence
How Open Models Affect User Control, Cost and Choice
A practical comparison of open access, open weights and open-source AI, examining hosted APIs, managed open-model services and self-hosting across data control, licensing, cost, operations, security updates and portability, with eight questions for decision-makers.
Artificial Intelligence
What Actually Makes AI Expensive?
From final training runs to HBM, advanced packaging, networking, depreciation, utilisation, inference scheduling and customer total cost of ownership, this article maps where AI spending goes and why a lower unit price does not guarantee lower total expenditure.