Why AI hardware advice needs a safety check
An AI that can't see your workbench is giving confident advice about a world it can't check. That's fine for code review. It's dangerous for lithium batteries.
The incident
We asked a text chatbot whether a motor could run straight off a battery. It said yes — confident, plausible, wrong. Direct battery hookup can spark on contact and cook the driver.
AI can die and come back. You can't. That asymmetry is why the LivePair machine exists — an assistant that can actually see the wiring is a different product than one guessing from text.
Rules that survive bad advice
One rule from our build is non-negotiable: **never connect a gear motor directly to a battery** — power it through a charger/controller per its manual. The full wiring guide is on the hardware page.
For any physical build, treat AI output as a draft — check it against the manufacturer manual, and let the assistant see the actual work when you can.
Uncertainty you can see
LivePair's HUD exposes the agent's confidence live — when it's extrapolating instead of reading, that shows. For physical work, visible uncertainty is a safety feature, not a gimmick.