Opus 4.7 is better at using file system-based memory. It remembers important notes across long, multi-session work, and uses them to move on to new tasks that, as a result, need less up-front context.
大多数人认为AI模型在长对话中会逐渐'忘记'早期信息，需要不断重复上下文。但作者认为Claude Opus 4.7能够跨会话记忆重要信息，这挑战了人们对AI短期记忆局限的认知。这种持久记忆能力意味着AI可以真正进行长期项目，而不需要用户不断重复提供背景信息。
In a 1-million-token context, V4-Pro uses only 27% of the computing power required by its previous model, V3.2, while cutting memory use to 10%.
大多数人认为AI模型处理更长上下文必然需要更多计算资源，但作者认为DeepSeek V4通过创新架构实现了惊人的效率提升，大幅降低了计算和内存需求。这一反直觉的发现挑战了'长上下文等于高成本'的行业认知。
The inbox becomes the agent's memory, without needing a separate database or vector store.
大多数人认为AI代理需要专门的数据库或向量存储来维护状态和记忆，但作者提出一个颠覆性观点：电子邮件收件箱本身可以作为代理的内存系统，这挑战了构建AI代理时需要复杂后端存储的行业共识，暗示电子邮件可能是一种未被充分利用的状态管理工具。
Large language models live in a similar perpetual present. They emerge from training with vast knowledge frozen into their parameters but they cannot form new memories – cannot update their parameters in response to new experience.
这个观点挑战了我们对AI学习能力的传统认知。LLMs虽然拥有大量知识，却无法像人类一样形成新记忆，这揭示了当前AI系统的根本局限性。作者通过《记忆碎片》电影中的失忆症患者类比，生动地展示了当前AI系统的'永恒现在'状态，这是一个反直觉的深刻洞见。
Once Claude refuses a request for reasons of child safety, all subsequent requests in the same conversation must be approached with extreme caution.
这一指令暗示Claude具有某种'记忆'或'状态追踪'能力，即使拒绝请求后仍会记住之前的拒绝。这与传统AI模型的无状态特性形成鲜明对比，表明Claude可能具有某种会话上下文记忆机制，这一反直觉特性可能被开发者忽视。
Agent harnesses dominate agent building and tie intimately to memory.
令人惊讶的是：代理工具(harnesses)已成为构建AI代理的主导方式，并且与记忆系统紧密相连。这表明AI代理的发展方向已经从单一功能转向了具有记忆能力的复杂系统，这种转变可能彻底改变人机交互模式。
After each round the model generated a memory file, criticized its own results, and fed those observations into the next round.
令人惊讶的是：M2.7模型能够生成自己的记忆文件，批判自己的结果，并将这些观察反馈到下一轮训练中。这种自我反思和持续学习的能力类似于人类的元认知过程，展示了AI系统越来越接近人类的自我评估和改进能力。
Unlike traditional GPU-centric systems, MegaTrain stores parameters and optimizer states in host memory (CPU memory) and treats GPUs as transient compute engines.
令人惊讶的是：这项研究彻底颠覆了传统GPU训练范式，将百亿参数模型的训练重心从GPU转移到CPU内存，这打破了人们对GPU作为AI训练核心的固有认知。这种'GPU仅作为计算引擎'的理念可能重新定义大模型训练的基础架构。
OpenClaw update gives Claws light, REM, and deep 'sleep' cycles to consolidate short-term memories into long-term ones.
令人惊讶的是：AI助手现在被设计有类似人类的睡眠周期，包括轻度睡眠、REM睡眠和深度睡眠，用于将短期记忆巩固为长期记忆。这一设计模仿了人类记忆形成的过程，展示了AI系统设计中越来越复杂的生物模拟元素。
existing methods typically attempt to address both issues simultaneously using a single paradigm
大多数人认为解决长时程LLM代理问题应该采用统一的方法同时处理全局进度和局部可行性，但作者认为这两种挑战本质上是不同的：一个依赖模糊语义规划，另一个需要严格逻辑约束和状态验证。这种分离的观点挑战了当前AI研究的主流范式。
内置视频和音乐生成 记忆系统学会了"做梦"
大多数人认为AI的记忆系统只是简单的数据存储和检索功能，但作者暗示OpenClaw的记忆系统已经发展出类似人类'做梦'的能力，这是一种具有创造性和联想性的高级认知功能，挑战了人们对AI记忆系统的传统认知。
Robert Lender blogpost about generated fake imagery to manipulate or suggest memories of people around you. A dinner, a Santa visit. Or leaving your (grand)children faked photos without being marked as such (look dad was on Greenland!) Interesting thought experiment. Bc memories are not fixed, and iterated upon with every retelling. Influencing that retelling is a given possibility. Turbocharged gaslighting too.
Chief Technology Officer of Agentforce, pointed out that when given more than eight instructions, the models begin omitting directives—a serious flaw for precision-dependent business tasks.
Whut? AI-so-human! Vgl 8-bits-schuifregister metafoor. [[Korte termijngeheugen 7 dingen 30 secs 20250630104247]] Is there a chunking style work-around? Where does this originate, token limit, bite sizes?
Facebook AI. (2021, July 16). We’ve built and open-sourced BlenderBot 2.0, the first #chatbot that can store and access long-term memory, search the internet for timely information, and converse intelligently on nearly any topic. It’s a significant advancement in conversational AI. https://t.co/H17Dk6m1Vx https://t.co/0BC5oQMEck [Tweet]. @facebookai. https://twitter.com/facebookai/status/1416029884179271684