Coding Tests Are Irrelevant: Why It’s Time for a New Approach

The traditional coding test, once a hallmark of technical interviews, is quickly losing its relevance in today’s AI-driven world. While these tests were designed to evaluate a candidate’s problem-solving skills and technical abilities, they’ve become outdated and, frankly, a poor measure of real-world talent. As someone who’s taken my fair share of these tests, and seen how irrelevant they can be, especially for management roles, it’s time we rethink the process entirely.

The Misalignment of Coding Tests and Job Requirements

One of the most glaring examples of this irrelevance is when companies use coding tests for roles that don’t even involve coding. For instance, why give a senior manager or principal engineer a coding test when the role requires no hands-on coding? Management positions should be about assessing leadership, decision-making, and how effectively someone can guide teams, not whether they can invert a binary tree.

Yet, we see these tests pop up in interviews for senior roles, where what really matters is how well a candidate manages people, budgets, and projects.

AI’s Role in Modern Development

The best coding test I’ve ever done wasn’t one where I had to solve abstract algorithmic puzzles, but one where I was asked to show that I could solve a challenging problem using AI. I had to fix a bug in an existing codebase, add a new feature, and submit not only the code but also a transcript of how I used AI tools to do it.

This was 100% relevant to how I think we should be using AI in development. In a world where AI is becoming a key part of every developer's toolkit, this test reflected the real-world skills that matter. It evaluated my ability to collaborate with AI, to debug, to implement features in existing code, all things that are far more important than memorising algorithms or solving puzzles under pressure.

This test wasn’t just relevant; it was practical. It mimicked how modern engineers work, relying on AI to optimise and automate tasks. We need more of this kind of assessment.

Coding Tests Without Resources Miss the Point

Another issue with traditional coding tests is their artificial constraints. Tests that forbid the use of online resources are unrealistic and fail to reflect how most engineers actually write code. In reality, no one works in a vacuum. Good engineers are resourceful. They know how to find solutions, debug issues, and optimise their work using the vast resources available online.

A coding test that prevents you from referring to documentation or using Google doesn’t test real-world skills; it tests your ability to operate under artificial conditions. This is a completely backward approach. The most successful developers are those who know how to solve problems efficiently, often by leveraging online resources. Penalising them for doing what they’d do on the job makes no sense.

The Future: Real-World Problem Solving and AI Collaboration

We need to move past coding tests as they currently exist. They’re irrelevant, especially when AI can solve them faster and more accurately than humans. What’s more important in today’s landscape is how developers can work with AI, not against it. The best candidates are not the ones who can write code in isolation but those who can integrate AI tools into their workflow, troubleshoot problems, and collaborate effectively.

The future of coding tests should reflect this reality. Companies should assess a candidate’s ability to solve real-world problems, work on existing codebases, and use AI tools to augment their productivity. Instead of abstract algorithmic problems, we should be giving candidates tasks that are directly related to the work they’ll be doing, like debugging, feature implementation, and optimisation, preferably with AI assistance.

A Call for Change

If we don’t evolve the way we assess technical talent, we risk continuing to hire based on irrelevant measures. Coding tests should not be about solving esoteric puzzles. They should be about how well a candidate can solve real problems in a way that reflects the way we work today.

The industry needs to embrace the tools we have and stop pretending that the coding interview of the past is still relevant. AI has changed the landscape, and our interview processes need to catch up.

In the end, we should be evaluating candidates on the skills that matter in a modern development environment: critical thinking, problem-solving, and the ability to use AI to get things done. Anything else is just noise.

See:

• https://theresanaiforthat.com/s/leetcode/
• LeetCode Helper: AI-powered Interview Prep & Problem ...