Felix Rieseberg, recruitment manager at Anthropic, said he has completely stopped programming tests when recruiting because AI is doing better than humans.

Rieseberg shared the change in the recruitment process on his blog over the weekend: "Since October, I officially killed the traditional coding interview. As recruitment manager at Anthropic, I stopped asking candidates to do them."

The reason, he says, is that the company's Claude models are now better at writing code than humans, making the technical test no longer necessary. When Claude demonstrated outstanding ability to grasp and process code, asking candidates to also do this type of test does not bring practical meaning in assessing capacity.

Instead, Rieseberg is looking for new methods and directions to recruit and evaluate young engineers. "I'm probably not the only manager struggling to find ways to interview people who know how to leverage AI to create effective products. I don't have the optimal solution yet, but I'm reviewing competency profiles to look for taste, empathy, and principled thinking," he said.

Anthropic's management also advises young engineers to focus on making a real impact, spending time learning about the company and the people working there. He appreciates candidates who proactively create change, such as building a prototype using AI on their own over the weekend to demonstrate their ability and problem-solving mindset.

For young engineers, Rieseberg conveys the message that "the desire to contribute and create value" now carries much greater weight than mere seniority or theoretical experience.

The fact that AI is doing a good job of programming is forcing the technology industry to reconsider its criteria for recruiting software engineers. Instead of testing limited-time coding, many businesses tend to evaluate the ability to identify problems, design solutions, test and detect flaws in AI-generated code.

A survey by Karat, a company providing interview and technical competency assessment services for businesses, of about 400 technical leaders in the US, India, and China in 2026 showed that 71% said AI makes assessing candidates' technical capabilities more difficult. Meanwhile, CoderPad's survey also noted that system design, debugging and AI collaboration skills are being valued more.

Huy Duc summary