I disagree. The amount of slop I need to code review has only increased, and the quality of the models doesn’t seem to be helping.
It still takes a good engineer to filter out what is slop and what isn’t. Ultimately that human problem will still require somebody to say no.