"A human will check it" is quietly carrying AI engineering demos
burhop · x · 2026-10-12
Writing on AICAD (AI-generated engineering design), the author identifies a blind spot in current demos: the AI revises a machined bracket, prepares the manufacturing program and writes a report in minutes — but when asked how we know it's correct, the answer is always "a human will check it."
That phrase introduces a person whose time, tools, information and responsibilities never appear in the demo. Which human? Are they checking geometry, machining instructions, material specs, or authorization? Can they trace the result back to the customer's actual request, and how long does it take? Until that part is shown, the productivity claim is unresolved: if your productivity story stops before approval, it may just be measuring someone else's future workload.
The post recalls that validation processes long predate chatbots — e.g. Trinity Forge's public quality manual requires CNC programs to be validated before release — and argues AI tools must incorporate this checking layer rather than bypass it.
More from coding & agent
- Logit-based decision models batch naturally: routing, scoring, yes/no in one forward pass — LysandreJik · 2026-10-12
- Hugging Face's transformers serve can turn 250 LLMs & VLMs into logit-based decision models — LysandreJik · 2026-10-12
- How One Dev Rents a 64-Core, 256GB Box for $800/Month to Run 30 Parallel AI Subagents — TheZachMueller · 2026-10-12
- NYC coffee shop spot check: only 6 of 22 laptops used AI, all ChatGPT — thedealdirector · 2026-10-12
- Beyond hallucinations: Reddit discusses how to ensure agents' output completeness — KDavidP1987 · 2026-10-12
- LocalLightChat: a lightweight Open WebUI alternative now with Agents, MCP and RAG — PromptInjection_ · 2026-10-12