问HN：人工智能开发工具准备好投入生产了吗？你们遇到了哪些限制？
大家好，
我注意到，基于人工智能的开发工具和服务在快速原型制作和加速应用开发方面具有显著潜力。然而，它们也伴随着一系列开发者和公民开发者/构建者常常遇到的限制和痛点。
例如：
- 人工智能的幻觉和不准确的代码生成导致大量手动修复。
- 供应商锁定使得数据和代码的导出变得困难。
- 定制化有限和组件僵化限制了独特功能的实现。
- 较差的调试工具妨碍了复杂问题的排查。
- 可扩展性问题影响生产就绪的应用程序。
- 不可预测的定价让人对人工智能和计算成本感到沮丧。
- 薄弱的协作和版本控制使团队工作流程复杂化。
你们也同意这些限制吗？
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?