提问 HN：AI 开发工具准备好投入生产了吗？你看到了哪些局限性？
3 分•作者: skykarthick•5 个月前
大家好，
我注意到，由人工智能驱动的开发工具和服务在快速原型设计和加速应用开发方面展现出巨大潜力。但与此同时，它们也带来了一系列局限性和痛点，开发人员和“公民开发者/构建者”们经常会遇到这些问题。
例如：
* 人工智能的“幻觉”和不准确的代码生成，导致需要大量手动修复。
* 厂商锁定使得数据和代码导出变得困难。
* 有限的自定义选项和僵化的组件限制了独特功能的实现。
* 糟糕的调试工具阻碍了复杂的故障排除。
* 可扩展性问题影响了生产就绪型应用程序。
* 不可预测的定价导致对人工智能和计算成本的担忧。
* 薄弱的协作和版本控制使团队工作流程复杂化。
您也认同这些局限性吗？
查看原文
Hello Everyone,<p>I see that AI-powered development tools and services are offering significant potential for rapid prototyping and faster application development. On other the side, it comes with a range of limitations and pain points that developers and citizen developers/builders frequently encounter.<p>For example:<p>- AI hallucinations and inaccurate code generation that lead to extensive manual fixes.<p>- Vendor lock-in makes data and code export difficult.<p>- Limited customisation and rigid components restrict unique features.<p>- Poor debugging tools hinder complex troubleshooting.<p>- Scalability issues impact production-ready applications.<p>- Unpredictable pricing creates frustration over AI and compute costs.<p>- Weak collaboration and version control complicate team workflows.<p>Do you agree on those limitations as well?