OpenAI 研究员 Ambrosino：AI 能写代码，但决定做什么仍难
精选理由
OpenAI 的 Andrew Ambrosino 聊了个真问题：模型能写代码，但决定做哪个功能还是得靠人，举了不少实际例子。
OpenAI 的 Andrew Ambrosino 在访谈中指出，AI 模型写代码已经不难，难点在于决定该做什么产品：如何权衡用户反馈、如何排定功能优先级。他以软件里临时加的 helper 工具为例，说明这些辅助会随模型能力提升而被原生替代。整段内容围绕 AI 做产品决策的局限展开。
原文 · The Information
AI can write the code. Deciding what to build is still a challenge.
OpenAI’s Andrew Ambrosino explains why maintaining a large product—from weighing feedback to prioritizing features—remains difficult for AI models.
“But it's always these arcs of things where like, hey, we're going to introduce this little helper here in the software. And at some point we'll be able to retire it because the model will be able to do that natively."
@ajambrosino
