I was grinding my way through the forest up a steep, switchback trail. It was one of those long, lung-burning climbs where your legs ache, your breath turns rhythmic, and every corner you clear without stopping feels like a small triumph.

Then, from behind, came a quiet hum.

A younger rider passed me, upright, relaxed, gliding effortlessly on an e-bike. We nodded, two solo riders enjoying the same trail. As he disappeared up the hill, I caught myself thinking: we’re not climbing the same mountain. We’ll both reach the top, sure. But what’s happening inside us is entirely different.

It wasn’t resentment I felt, more like recognition, a moment of clarity that our goals were simply different.

What Are We Actually Climbing For?

For me, the climb is the reward. The effort is the point. It’s the friction that strengthens, the rhythm that quiets the mind, the slow conversion of discomfort into confidence. Every steep section conquered whispers the same thing: you got this.

For him, that day, maybe the goal was range, to see more trails, more vistas, to experience more in less time.

And neither of us was wrong. We were simply playing different games where one focused on growth through effort, and the other on expansion through assistance. That realization stuck with me, because the same dynamic is playing out everywhere today, not on trails, but on keyboards.

The New “E-Bike” for the Mind

The frontier of efficiency has shifted. Writers now draft with AI. Developers use it to generate code. Some still prefer the manual grind, crafting each sentence, debugging every loop line by line.

And again, the same question arises: what are we climbing for?

When you write or code entirely by hand, the work changes you. You learn to sit with the blank page, the compiler errors, and the long silences where ideas have to form from scratch. You build clarity, a sense of authorship, and a depth of understanding that can’t be outsourced.

When you use AI, the work shifts. You move faster. You operate at a higher level of abstraction with more time spent shaping than forging, curating than composing. Different routes. Different kinds of flow. In one path, the self expands through friction; in the other, through reach. However, without strong fundamentals, we risk falling into the competence trap.

The Competence Trap

Just because the frontier of efficiency in our field has moved does not mean we can abandon the fundamentals. Systems thinking has a concept called “shifting the burden to the intervenor.” A helpful intervention can make the underlying problem easier to ignore, and your capability to solve it yourself gradually diminishes. You become dependent, and when the intervention fails or is unavailable, you are worse off than before it arrived.

There is a critical difference between engineering with AI and shifting the burden to AI. Engineering with AI involves directing, validating, and developing judgment about outputs. You amplify competence. Shifting the burden to AI can erode our grasp of the fundamentals, leaving us unable to judge the quality of what’s produced. You move forward, but you’re no longer piloting. Errors slip through. Bad code still runs. Skills atrophy. That is not growth. It is accumulated fragility waiting for an edge case you can’t handle.

Working With AI, Not Under It

So when should you take the e-bike, and when should you grind? The answer isn’t binary. AI isn’t the enemy of growth, nor is it a shortcut to mastery. It’s a tool whose value depends on what you’re optimizing for in the moment.

Output mode is when the goal is delivery, shipping, rapid prototyping, or otherwise reaching a destination on time. In that context, AI is the pragmatic choice. Using it to prototype or draft isn’t cheating, it’s the new frontier of engineering efficiency.

Growth mode is when you are deliberately building new capabilities. Sometimes you need friction, manual work that reveals shallow understanding and strengthens judgment for future AI use. Growth mode is not necessarily about avoiding AI. It is about consciously working at the edge of your capability, whether you’re doing pure or impure engineering. Examples:

• An experienced developer using AI to explore architecture options is in growth mode with AI.

• Debugging a complex algorithm manually to reinforce fundamentals is growth mode without AI.

• Copying AI-generated code you cannot explain is neither growth nor effective output; it is the competence trap.

Some people resist AI entirely while others rely on it for everything. Both avoid the real decision, what am I actually trying to build right now, capability or output?

Different Mountains, Different Rewards

I reached the top of that trail long after the e-bike rider had gone. He had probably seen twice the terrain I had. Standing there, heart pounding, sweat cooling in the breeze, I felt something I valued, the quiet satisfaction of having done the hard thing the hard way. That was my chosen reward, the internal currency of effort and accomplishment.

And yet, I also understood the appeal of his ride. The freedom, the range, the efficiency. He had gained something too, just a different kind of reward. We were in fact climbing the same mountain, but that day, our goals were different.

Recognize your goals and be intentional about how you use AI to assist, not to intervene.

The Choice in Action

This essay itself illustrates the distinction, fitting somewhere between those two climbs. I began with an outline of ideas stemming from that moment on the trail, some thoughts about flow, effort, reward and my personal experiences developing software with AI. With that in hand I shaped those thoughts into well formed paragraphs with the help of AI. I optimized for ‘time to first draft’, not for the long, lonely muscle-building work of writing every sentence by hand. Here, I chose the e-bike.