Computing

Traces the evolution from Transformer to ChatGPT, arguing that foundation models' core features are few-shot learning without fine-tuning and conversational UI, with large-scale training being the means rather than the goal.

对比分析Quora旗下AI平台Poe与ChatGPT的优劣，分享多模型对比、多Bot协作等功能体验，以及替代ChatGPT核心功能的实际使用感受。

Why I switched from ChatGPT to Poe - a comparative analysis of Quora's AI platform that integrates GPT, Claude, Llama and more, with multi-bot collaboration and prompt comparison features.

展示Claude V3.5长上下文和不偷懒特性的创意实验，以"剑宗的师叔气功好，气宗的徒儿剑法高"为题，让AI撰写了一篇13000字的学术论文，探讨金庸武侠中华山派的能力发展机制。

5分钟用AI给全天相机加上云量识别功能：利用GPT-4V理解云量，让AI自动编写调用API的代码，展示AI时代开发效率的革命性提升。

构想一个AI知识引擎：自动收集会议和日常对话、提取有价值的知识并索引、在输出时及时提醒。三大挑战是数据采集、知识提取和实时参与。

Envisioning an AI knowledge engine that records conversations, extracts insights, and surfaces relevant knowledge when needed. Three key challenges: data collection, knowledge extraction, and real-time participation.

分享开发视频语音识别工具的技术心得：YouTube下载OAuth认证、异步任务API设计、Python并发挑战、GPT偷懒问题和iOS Shortcut集成。

Technical lessons from building an AI video transcription tool: YouTube OAuth2 authentication, async task-based APIs, Python concurrency patterns, GPT laziness issues, and iOS Shortcuts integration.

提出AI辅助编程的核心从数据结构和算法转向注释和DocString的质量。类比管理AI与管理人类团队，强调了解能力边界、任务分解、委托和质量检查等开发经理式的思维方式。