Unverified60% confidenceFactExact time
When AI models generate code, they lack persistent memory of the overall architecture and tend toward local patching rather than global refactoring in each iteration, causing cyclomatic complexity and coupling to continuously rise.
2
Sources
60%
Confidence
Long-term
Relevance
7/6/2026
First Seen
Sources
Harness Engineering: The Core Methodology for Advanced AI Programming
bilibili12点就睡的林同学7/3/2026
Related Claims
UnverifiedAI在小型重构任务上表现出色，但在处理高度耦合、上下文庞大的遗留系统时会暴露理解不完整、细节丢失的短板77% similarVerifiedIn the process of repeatedly patching bugs, AI introduces redundant logic, duplicate code, and contradictory implementations, causing technical debt to accumulate far faster than in traditional development.77% similarUnverifiedAI 生成代码产生混乱的根本原因是LLM是无状态的，不存在跨会话的持久化记忆，与软件工程要求的强一致性存在结构性矛盾77% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/113837API
curl https://kongchang.com/api/v1/knowledge/claims/113837MCP
get_claim(id=113837)