Unverified50% confidenceOpinionExact time
Current large language models can, at best, substitute for only the code-writing stage of software development
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
7/2/2026
First Seen
Valid until: 9/30/2026
Sources
Related Claims
Unverified大语言模型在代码生成时可能混淆不同版本的API，或遗漏实际编译所需的import语句75% similarPartially Verified大型语言模型不仅理解编程语言语法，还掌握了大量设计模式、最佳实践和常见问题解决方案73% similarUnverified当前主流大语言模型在处理纯文本时表现优异，但对表格、流程图、截图等非文本格式的理解能力存在显著瓶颈72% similarUnverified当前大语言模型在处理长上下文、维护复杂状态和保证代码模块间一致性方面仍存在显著局限72% similarUnverified大型语言模型本身只能生成文本，无法直接操作文件系统或网络，实际执行能力依赖开发框架预先注册的工具函数71% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/50852API
curl https://kongchang.com/api/v1/knowledge/claims/50852MCP
get_claim(id=50852)