ChatGPT solves FizzBuzz in six languages with perfect documentation. Traditional coding interviews are now testing whether candidates can resist the urge to use tools they will use every day on the job.
AI does not equalize skill. It amplifies existing ability. A senior engineer with AI ships faster and at higher quality. A junior engineer with AI ships plausible-looking code with subtle bugs. The gap widens, not narrows.
This changes what interviews should measure.
The Shift
Traditional interviews tested whether someone could solve problems from memory. The relevant skill is now whether someone can solve problems with AI -- and recognize when the AI got it wrong.
The specific capabilities that matter:
Context management. Can they feed the AI the right information at the right time? Engineers who iterate endlessly without improving their context do not scale. This is the new literacy.
Quality recognition. Can they look at AI-generated code and immediately spot subtle bugs, performance issues, and maintainability concerns? Anyone can ask AI to "make it faster." Recognizing when it actually made it slower requires deep understanding.
System thinking. AI excels at local optimization. It struggles with global architecture. The developers who succeed see the bigger picture and guide AI toward solutions that fit the entire system, not just the immediate function.
The Unsolved Economics Problem
Watching someone work with AI for 2-3 hours produces incredible signal. It is also expensive. You cannot do this for every candidate.
Phone screens that used to filter most candidates are now useless -- AI can answer any technical question. We need new filtering mechanisms that work at scale. Nobody has cracked this yet.
The companies still running traditional coding challenges are hiring for 2015. The companies running AI-paired work sessions are getting better signal but cannot scale the process. The solution likely involves asynchronous AI-paired take-home projects with automated signal extraction, but that tooling does not exist in production form.
This is an open problem. The hiring process that replaces coding interviews has not been invented yet.