# AI Will Make You "Look" Competent. That's the Problem. #AI *Last Updated: March, 2026* "You don't need to learn X in the age of AI" is one of the most viral pieces of advice floating around right now. It's also one of the most dangerous. To be clear: this isn't a nostalgia argument. AI is real, and ignoring it is its own kind of mistake. But there's a trap hidden in that advice that will hollow out your capabilities in slow motion, and it won't announce itself while it's happening. Think about following GPS every day without ever building a mental map of the city. You navigate fine, right up until the signal drops or you're somewhere the map doesn't cover. AI-assisted work on unfamiliar terrain operates the same way. Fluent until it isn't. Competent-looking until the moment it matters. The fix isn't to avoid AI. It's to use it differently depending on which side of a line you're on. ## The First Thing to Do: Sort Your Work Before you decide how to use AI on any given task, sort it into one of two buckets. **Mode 1:** Terrain where you already have honed skills: work you understand deeply, where you can tell good output from mediocre output without thinking hard about it. **Mode 2:** Terrain that is genuinely new to you: work where you're still developing the ability to evaluate, not just execute. The self-diagnostic is simple. For any task you're about to do with AI, ask: if I had to explain why the output is good or bad, could I? If yes, you're in Mode 1. If no, you're in Mode 2. Same tool. Different playbook. Think of an experienced cyclist who uses gears to maintain cadence and manage climbs, versus a beginner who just mashes through them without reading the terrain. The gears are identical. What they do for your development isn't. ## Mode 1: Compress Time, Compound Leverage On familiar terrain, use AI aggressively. This is not optional advice. If you've genuinely developed a skill, using AI to execute faster frees you to operate at higher levels of abstraction, where your judgment actually lives. The leverage compounds because your judgment acts as a quality filter. An experienced photographer using AI-powered editing tools moves through adjustments in minutes (contrast, color grading, cropping) because they know exactly what serves the image. Someone still learning photography applies the same tools and can't tell whether the result is better or just different. A few things that help here. First, be honest about which skills are actually honed: it's a short list, not a long one. Second, use AI to compress execution time so you're spending cognitive energy on the decisions that require your pattern recognition, not the ones that don't. Third, actively seek AI outputs that surprise you. That's where your judgment gets sharpened. That's where you find the edges of your own mental model. And track where your interventions improve the output. That's your leverage map, and it grows every time you use it. ## Mode 2: Use AI to Find the Gap, Not Fill It On new terrain, the instinct to use AI as an answer machine is exactly wrong. The short burst of productivity is real, but it's borrowed. Without a deliberate learning approach underneath it, the productivity fades and the skill never forms. Physical therapists know that a muscle passively moved heals slower than one that actively strains against resistance. When you use AI to do the work on terrain you don't yet understand, you're in passive rehab. When you use AI to surface what you don't understand, you're doing the actual work. Three things are required, and none of them can be outsourced: **1. Discover what needs to be done.** Use AI output as a starting point for a post-mortem, not a deliverable. Before you accept what it gives you, ask: what would I have to believe to think this is good? What's missing? What did it skip over? The answers to those questions are the map of your ignorance, which is exactly what you need. **2. Articulate the why.** Generate multiple AI outputs on the same problem and force yourself to rank them with reasons. Not "this one feels better." Reasons. Why is one framing stronger? Why does one solution fail? This is how you surface your own criteria instead of just consuming answers. You are not evaluating AI here. You are developing judgment. **3. Cultivate an opinion for what good looks like.** Before you prompt AI on something unfamiliar, spend 20 minutes attempting it yourself. Not to produce the right answer, but to prime the learning. Struggle activates something that passive consumption doesn't. When you then see the AI's output, you'll see it differently. You'll have something to compare it against. ## The Trap You Won't Notice Until It's Too Late Here's the thing nobody talks about: AI-assisted output on unfamiliar terrain generates enough positive feedback to feel like learning. The work looks good. People respond well to it. You feel productive. And slowly, without a deliberate approach, you stop developing the underlying skill. I've done this. It feels exactly like competence until something breaks the loop: a harder problem, a smarter room, a moment when the AI can't tell you what question to ask. It's the answer key trap. Reading the solution before solving the problem feels like understanding. You've seen the right answer. The understanding isn't there. The test always finds this out. The behavioral flag is simple: if you can produce the output but can't critique or defend it, slow down. That's not productivity. That's drift. The moment you notice it is the moment to do the opposite of what feels efficient. ## What AI Actually Changes AI doesn't change what learning requires. It changes how visible your gaps become and how fast you can develop once you decide to. The real upside isn't skipping the learning. It's compressing the feedback loop for the learning you actually commit to. This week, pick one unfamiliar thing and use AI to find the gap, not fill it. That's where the learning starts.