This feels like the same principle as “measure twice, cut once”

WTF! The guy that finishes everybody with no problems is on the wrong and the one that is pushing AI into everybody is right? What kind of conclusion is that?

Absolutely silly premise. Continuing the author’s analogy, how would pacing oneself in one marathon improve performance in the race tomorrow? What about next week (a race of indeterminate length)? Or next month (turns out will be a triathlon, but you don’t know it yet)?

It doesn’t, and the “races” that engineering teams run are more like those latter examples than they are like performance during a single marathon.