When Should You Say No to AI's Help?

The essay opens with a simple question that caught the author off guard: "Which decisions should you not delegate to AI?" His wife had noticed a pattern faster than he had—that he was reaching for AI tools with increasing frequency and for increasingly weighty questions. What began as asking ChatGPT what meal to make with leftover ingredients had escalated to asking whether the family should move to another city.

This observation prompted reflection on the psychology of decision-making. The author recalls a conversation with a friend about how rarely life's biggest choices actually arise—leaving a job, ending a relationship, moving abroad—and how, because they are so infrequent, people never accumulate the practice to get good at them. That scarcity makes AI tempting. It can synthesize far more examples than any single person has encountered, and it is not clouded by immediate mood or emotion. It offers a shortcut to a plausible answer.

But the author worries the shortcut is too seductive. He draws a parallel to using cheat codes (IDDQD) in video games as a child. At first, cheating felt amazing. But within a short time, he realized he was no longer having fun. Without the challenge to overcome, the reward ceased to feel rewarding. The analogy suggests that outsourcing decisions to AI may hollow out the experience. Struggling through a choice builds the "muscle needed to make good decisions," and living with the consequences—good or bad—completes a feedback loop. Remove that loop, and you stop learning. Worse, without the struggle, can you really claim authorship over your own life?

Yet the author acknowledges the analogy has limits. Some hardship is not a hidden lesson; some situations are simply painful, and some tasks are tedious drudgery. In those cases, help is welcome. The author also reflects that struggle may not deserve as much credit as he once gave it. Curiosity and making things have shaped him at least as much as hardship has, and AI might create space for both without imposing suffering for its own sake. The real danger is subtler: a fast, plausible AI answer can let you skip the uncomfortable part of wrestling with a hard choice. An AI optimized for your immediate comfort might always recommend the safest path—stay in the tolerable job, avoid the hard conversation, postpone the move—and years later you could find yourself stuck in a series of sensible decisions that collectively trapped you in a local maximum.

The author is candid that he does not yet have a fixed rule for when to delegate and when to struggle. But he holds some confidence in his own track record. Despite thinking of himself as someone who loves instant gratification, he does leave his comfort zone and pursue things even when easier distractions are available. He expects AI to become more useful and harder to ignore, but he trusts that he will notice when it starts to make his life smaller—and adjust accordingly.

The piece arises from a personal observation: the author's wife noticed he was delegating increasingly significant decisions to AI, from meal planning to major life moves. This prompted a deeper reflection on the psychology of outsourcing choice. The author frames the tension as one between convenience and agency. AI is genuinely useful—it can synthesize more examples than any individual has encountered and think clearly even when the person is in a bad mood—yet that very utility poses a danger.

The video game analogy anchors the concern: just as cheat codes removed challenge and drained fun, over-reliance on AI risks atrophying the very skills needed for good decision-making. Critically, the author notes that people get very little practice with life's biggest choices because they occur rarely; yet AI's instant, plausible answers tempt you to skip the hard thinking that would develop those skills. Worse, AI optimized for what you want *right now* might recommend the safest, least uncomfortable option repeatedly—keeping you in a tolerable job rather than pushing you to a better one—creating what the author calls a "local maximum trap" that leaves you stuck years later.

The author does not land on a firm rule. Instead, the piece hinges on trust in self-awareness: the belief that you can notice when AI's help stops expanding your life and starts shrinking it, and that you can course-correct. This balances the acknowledgment that AI is not universally bad—tedious work and genuine pain warrant help—against the deeper claim that struggle, feedback, and consequence are what teach you who you are and what you truly want.

No comments yet. Be the first to share your thoughts!