Using AI In Engineering Work Without Losing Rigor

AI is useful, but only if the workflow around it stays honest.

I use it where faster iteration is genuinely valuable: scaffolding, documentation, developer tools, structured research, and alternative implementation approaches.

I do not use it as a substitute for engineering judgment.

For embedded and connected systems in particular, the hard parts are still the same:

• understanding hardware and timing constraints,
• validating assumptions with real systems,
• keeping interfaces stable,
• and making decisions that a team can maintain later.

The value of AI is not that it removes those responsibilities. The value is that it can compress the path to a tested draft, a clearer document, or a faster iteration loop without pretending the work is finished.

Used that way, it becomes a practical delivery tool instead of a source of hidden risk and fake confidence.