{"slug": "ai-will-not-stop-at-simple-work", "title": "AI will not stop at simple work", "summary": "AI is moving into complicated software work faster than many programmers realize, shifting the bottleneck from code production to specification and verification, according to a software engineer's analysis. The author argues that while current models struggle with complex legacy systems, they are rapidly improving at exploring codebases, running tests, and adjusting, making the human role increasingly about defining correct behavior, reviewing outputs, and taking responsibility.", "body_md": "After I wrote that [programming experience is no longer a\nmoat](/software/2026/07/22/programming-experience-is-no-longer-a-moat.html),\none reply kept appearing.\n\nAI can write simple code, people said, but when the system becomes complicated, you still need an experienced human.\n\nThis is a reasonable objection.\n\nIt also feels true today.\n\nAsk a model to build a small application and it moves quickly. Ask it to change billing rules across a ten-year-old system and it may wander into the woods. A human has to pull it back.\n\nThe mistake is treating that boundary as permanent.\n\nThe models that struggled with a complete feature last year can now explore a codebase, change several files, run tests, notice failures, and try again. They still get lost, but the forest keeps getting larger.\n\nThe replies focused on where the boundary is today. I was more interested in where it is moving.\n\nA complicated system is not complicated in one giant way. It is usually a pile of smaller decisions held together by constraints.\n\nThe payment must be retried, but never twice at the same time. Old customers keep their original price. The migration must work without downtime. Failed jobs must be safe to run again. This strange database column cannot be removed because a forgotten report still uses it.\n\nAn experienced programmer carries much of this context in their head.\n\nThe model does not.\n\nGive it a vague request and it fills the gaps with guesses. Give it the intended behaviour, the constraints, the edge cases, the relevant code, and a way to test the result, and something different happens.\n\nIt breaks the work down. It proposes a plan. It changes the code, runs the tests, reads the failures, and adjusts.\n\nIt will still make mistakes. Sometimes serious ones. I am not claiming that AI can already redesign every complicated system perfectly.\n\nI am saying that it is moving into complicated work faster than many programmers are updating their idea of what complicated means.\n\nThat shifts the bottleneck.\n\nThe difficult part becomes less about whether the machine can produce the code. It becomes whether the human can explain what correct means.\n\nThat is not a small task.\n\nWriting a good specification requires understanding the system, finding the hidden constraints, resolving contradictions, and choosing trade-offs. Then someone still has to review the result and take responsibility for it.\n\nThe human remains, but the work changes.\n\nMore of it becomes specification, judgment, verification, and ownership.\n\nIf you cannot describe the constraints, the model will invent them for you. If you can, you still have to check the result and put your name on it.\n\nThat is the part experienced programmers should be practising now.", "url": "https://wpnews.pro/news/ai-will-not-stop-at-simple-work", "canonical_source": "https://kyrylo.org/software/2026/07/23/ai-will-not-stop-at-simple-work.html", "published_at": "2026-07-23 11:17:27+00:00", "updated_at": "2026-07-23 11:42:29.502481+00:00", "lang": "en", "topics": ["artificial-intelligence", "ai-tools", "ai-agents", "developer-tools"], "entities": [], "alternates": {"html": "https://wpnews.pro/news/ai-will-not-stop-at-simple-work", "markdown": "https://wpnews.pro/news/ai-will-not-stop-at-simple-work.md", "text": "https://wpnews.pro/news/ai-will-not-stop-at-simple-work.txt", "jsonld": "https://wpnews.pro/news/ai-will-not-stop-at-simple-work.jsonld"}}