AIDan Luu argues AI coding agents frequently make errors that would get a human developer firedThe essay critiques LLM benchmarks for failing to capture real-world workflow friction.digg it1760123456789012345678901234567894 Sources, 138d ago, first seen 138d agoCombined views3.3K4 Sources, first seen 138d ago33 likesCombined views3.3K4 Sources, first seen 138d ago33 likes