1
Using GPT-4 to improve teaching and learning in Brazil
看GPT-4如何赋能巴西教育，提升教学与学习效率的真实案例
Improving teaching and learning in Brazil
2
Learning to reason with LLMs
OpenAI 分享如何训练大语言模型进行推理，揭示思维链背后的关键方法。
3
Pairing data with APIs to unlock customer value
乐天联手OpenAI，用API融合数据与AI，深度挖掘客户价值，商业实践案例值得关注。
Rakuten Pairs Data with AI to Unlock Customer Insights and Value
4
New compliance and administrative tools for ChatGPT Enterprise
ChatGPT企业版新增合规API、SCIM及GPT控制，助力大规模数据安全与用户访问管理。
Compliance API integrations, SCIM, and GPT controls to support compliance programs, data security, and user access at scale
5
The Newsroom AI Catalyst: a global program with WAN-IFRA
OpenAI联合世界报业协会发起全球新闻室AI催化剂项目，助力媒体创新。
6
Accelerating the development of life-saving treatments
OpenAI与Moderna合作，AI加速药物研发流程，展现AI在医疗领域的巨大潜力。
Accelerating the development of life-saving treatments.
7
GPT-4 API general availability and deprecation of older models in the Completions API
GPT-4 API正式开放，旧模型将逐步退役，开发者需留意迁移时间线
GPT-3.5 Turbo, DALL·E and Whisper APIs are also generally available, and we are releasing a deprecation plan for older models of the Completions API, …
8
Building agricultural database for farmers
OpenAI助力Digital Green构建农业数据库，展示AI在提升农民收入中的实际落地价值。
Digital Green uses OpenAI to increase farmer income.
9
Practices for Governing Agentic AI Systems
OpenAI发布代理AI系统治理实践，为构建安全可控的自主AI提供关键框架。
10
Our approach to AI safety
OpenAI官方阐述AI安全策略，从构建到部署全流程的安全管控框架。
Ensuring that AI systems are built, deployed, and used safely is critical to our mission.
11
GPT-4
OpenAI最新里程碑，多模态大模型GPT-4在专业学术基准上达到人类水平。
We’ve created GPT-4, the latest milestone in OpenAI’s effort in scaling up deep learning. GPT-4 is a large multimodal model (accepting image and text …
12
New AI classifier for indicating AI-written text
OpenAI官方推出AI文本检测器，精准区分人类与机器写作，为内容真实性护航
We’re launching a classifier trained to distinguish between AI-written and human-written text.
13
The power of continuous learning
OpenAI研究员Lilian Weng深入探讨持续学习在AI中的关键作用，揭示模型如何不断进化适应新任务与数据，干货满满。
Lilian Weng works on Applied AI Research at OpenAI.
14
DALL·E now available without waitlist
OpenAI宣布DALL·E取消等待名单，新用户可立即体验AI图像生成，安全性同步升级。
New users can start creating straight away. Lessons learned from deployment and improvements to our safety systems make wider availability possible.
15
Our approach to alignment research
OpenAI详解对齐研究策略：提升AI从人类反馈学习，目标打造能自动解决其余对齐难题的超级对齐系统。
We are improving our AI systems’ ability to learn from human feedback and to assist humans at evaluating AI. Our goal is to build a sufficiently align…
16
DALL·E now available in beta
OpenAI 图像生成工具 DALL·E 进入 beta 阶段，百万用户可免费试用，额外生成明码标价
We’ll invite 1 million people from our waitlist over the coming weeks. Users can create with DALL·E using free credits that refill every month, and bu…
17
AI-written critiques help humans notice flaws
OpenAI训练批评写作模型，让人类更容易发现AI总结的缺陷，揭示AI辅助监督AI的潜力
We trained “critique-writing” models to describe flaws in summaries. Human evaluators find flaws in summaries much more often when shown our model’s c…
18
Best practices for deploying language models
三大AI巨头联合发布的大语言模型部署最佳实践指南，行业必读。
Cohere, OpenAI, and AI21 Labs have developed a preliminary set of best practices applicable to any organization developing or deploying large language…
19
Measuring Goodhart’s law
OpenAI深度解析：当指标成为目标时如何避免失效，古德哈特定律在AI优化中的实践启示
Goodhart’s law famously says: “When a measure becomes a target, it ceases to be a good measure.” Although originally from economics, it’s something we…
20
Solving math word problems
OpenAI新系统解小学数学题，准确率追平9-12岁儿童！
We’ve trained a system that solves grade school math problems with nearly twice the accuracy of a fine-tuned GPT-3 model. It solves about 90% as many …