Have you ever felt unstoppable with an AI coding assistant at your fingertips?

The rush of watching lines of code appear almost telepathically can be exhilarating.

For a while, I thought AI tools were my superpower — until subtle cracks started to show in my own coding abilities.

The Seductive Speed (and the Hidden Cost)

Like many, I dove headfirst into AI-powered coding: Cursor AI, Claude Sonnet 3.7, and the works. It felt like magic at first. My output soared, and I felt like I’d gained a new superpower. But slowly, a nagging suspicion took root.

• The AI Pause: I’d catch myself stopping mid-thought to wait for an AI suggestion. My own problem-solving process seemed to stall.

• Syntax Amnesia: Basic constructs I’d known for years felt fuzzy around the edges. Why think it through if AI could fill in the blanks for me?

• Prompt Wrestling: At times, coaxing the right output from the AI took longer than if I’d just written the code myself.

It felt like I was moving faster, but was I really losing something crucial?

The “Self-Driving” Trap: Forgetting the Fundamentals

A recent article from Luciano Nooijen compared AI-driven coding to relying on Tesla’s Full Self-Driving: letting the car handle the mundane parts so often that, when you do need to take the wheel, everything feels off.

That analogy hit home. I realized that by automating the easy stuff — boilerplate code, basic loops, quick error handling — I was skipping my own practice reps. Over time, those reps build muscle memory and sharpen intuition.

The Interview That Exposed My Over-Reliance

Then came the coding interview that shattered my illusions. I was asked to write a simple function that returns True if two elements of a list sum to a target n, and False otherwise:

def sums_to(numbers: list[int], n: int) -> bool:

A year ago, I’d have breezed through that. But with AI doing most of my coding lately, I blanked. What should’ve taken a minute or two felt painfully slow. My thought process froze while I waited for an imaginary AI suggestion.

I didn’t complete the solution in time, and I lost the interview. That’s how badly my reliance on AI had eroded my confidence and speed.

For reference, here’s one straightforward solution:

def sums_to(numbers: list[int], n: int) -> bool:
    results = set()
    for el in numbers:
        if n - el in results:
            return True
        results.add(el)
    return False

In hindsight, it’s simple: track seen elements in a set, check if n - el exists, and return True if so. But at that moment—deprived of AI autopilot—I stumbled.

More Than Just Rustiness: The Real Risks

My interview hiccup served as a wake-up call, confirming what others had been warning me about:

• Security Blind Spots: Can I really vouch for code I didn’t deeply think through?

• Complexity Hurdles: AI struggles with edge cases, legacy systems, or niche frameworks; that’s where human expertise is vital.

• Skill Ceiling: If my top “skill” is prompting an AI, I risk capping my own potential or, worse, automating myself out of relevance.

Reclaiming Control: Using AI Without Losing Myself

Reading about others’ experiences didn’t make me rage-quit AI — it made me more strategic about how I use it. Now, I strive to be a pilot, not a passenger:

1. Active Reviewing: I don’t just accept AI suggestions at face value. I read them, refactor, and ensure I understand every line.

2. Valuing the Workout: I consciously write more fundamental code myself — even if it takes a bit longer. That’s practice.

3. The Airplane Test: If I had no internet, could I still get the job done? If the answer is no, time to put the AI aside and sharpen my own skills.

4. Strategic Use: I reserve AI for genuinely tedious tasks or for initial research on unfamiliar libraries. I’m still the one doing the thinking.

Rediscovering the Joy of Competence

The biggest upside? I’ve regained the fun in coding. There’s a unique satisfaction in wrestling with a problem and solving it through your own skills, not just prompting an AI. Speed is great, but competence feels better.

That quote rings true: AI was dulling my edge by making things too easy, and I almost forgot how much I enjoy the challenge of solving puzzles myself.

Over to You

Have you felt this friction between AI speed and your own skills? Are you rethinking how you use these tools?

I’d love to hear about your experiences.

If you have something to share, answer this email and tell your story of how AI impacts your coding muscles!