I just watched Apple's introductory agentic coding video for Xcode 26.3, and my takeaway is it adds a huge cognitive load to programming.
From my perspective, instead of figuring out how to do something and using that knowledge going forward on future projects, a programmer is reduced to acting as a babysitter to figure out if the agent did things correctly, and then spending a lot of effort reviewing the generated code to see if it is even correct.
More work, less knowledge—agentic coding is not the solution.
