the productivity gains from coding agents are not evenly distributed. they’re split along a k-shape: senior engineers are getting meaningfully more productive. junior engineers are, at best, treading water. at worst, they’re getting worse.
This is exactly what we should expect. You have to be experienced (as a coder, as a researcher, as a thinker) to discern hallucinations and fakery. Only experience will make that alarm bell ring in your head. Younger and less experienced coders/scholars/writers will uncritically accept whatever their preferred chatbot offers them, and that will all too often lead them into more and more profound errors.