Are coding interviews still relevant? AI cheating & the future of tech recruitment

Recently, X (formerly, Twitter) has been flooded with discussions about InterviewCoder.co, an AI tool designed to solve engineer recruitment challenges automatically. People call it an “invisible AI.” This tool helps candidates pass online coding interviews undetected. Imagine showing up to an online coding test, turning it on, and letting AI spit out the perfect solution instantly. No stress, no struggle, not even a slight preparation. If this is the new reality, it raises a critical question: What’s the point of traditional coding interviews anymore?

AI cheating or just adapting to the future of work?

Back in the day, cheating on coding interviews was such a hassle. The best we could do was sneak in a second monitor, Google solutions, or memorize common patterns. Since AI, cheating on coding assessments is on a whole other level. AI can generate complete, optimized solutions, explain them in detail, and perform better than many human candidates. Not only does AI know the answer, but it can also articulate the logic flawlessly and pass interviews better than we can.

The rise of AI in completing coding assessments raises an important question: Is this truly cheating, or is it simply the natural evolution of the industry? Netizens argue on X that traditional coding tests, like LeetCode-style challenges, were already a flawed measure of real-world ability. After all, engineers routinely use AI tools, like ChatGPT or Cursor, in their day-to-day jobs, be it to speed up generating some boilerplate components or generating some test case possibilities. Why should coding assessment be any different? But on the flip side, if AI can solve all the problems for us, then what skills are companies even testing for? Are they assessing genuine problem-solving ability, or just the candidate's capacity to leverage technology to game the system?

Recruiters are already starting to catch on, though. A solution that is "too perfect" can be a red flag. As a result, some companies are scrambling to implement sophisticated AI-detection measures, while others are beginning to completely rethink their entire tech interview process. Maybe it's time for a better way to hire.

The real problem: coding interviews were already broken

Let's be honest: coding interviews were deeply flawed long before AI came into the picture. Many companies still rely on an evaluation system that has little to do with the actual demands of an engineering role. You might be asked to implement a quicksort algorithm from scratch, but once you land the job, you're debugging APIs, reviewing pull requests, and dealing with system architecture. The disconnect is real.

Then, there’s the "speed coding" nonsense. Interviewers expect candidates to think, code, and explain their solution in minutes. In reality, real-world engineering work is a collaborative, iterative process that involves research and critical thinking. Now, with AI able to generate solutions instantly, the absurdity of these timed tests is even more obvious. Are we testing problem-solving skills or just measuring who can type the fastest under pressure? The need for a better hiring strategy has never been greater.

If not a coding assessment, then what?

If coding interviews in their current form are becoming pointless, what’s the alternative? The future of tech recruitment lies in assessments that mirror real-world work. Here’s what might actually work:

• Project-Based Assessments Instead of asking random algorithm problems, have candidates complete a real-world coding challenge. If you’re hiring a frontend engineer, ask them to build an actual UI component. If it’s for a backend role, give them a small API to design. This assesses the candidates’ actual skills instead of memorization.
• Live Coding with Explanation Move away from timed tests.. Instead of making candidates rush through code, let them explain their thought process. Give them time to think and discuss trade-offs. This shows how they approach problems rather than just whether they’ve memorized solutions.
• Pair Programming Engineers rarely work in isolation. So, why not assess them by pairing them with an interviewer to solve a problem together? This reveals communication skills, collaboration, and debugging ability, all things that matter far more than whether you can reverse a linked list on demand.
• System Design Interviews AI might be good at spitting out code, but it struggles with big-picture thinking. System design interviews force candidates to consider scalability, trade-offs, and architecture. This provides a much more robust test of a candidate's engineering maturity and strategic thinking—skills that AI cannot easily replicate.