Job interviews are getting complicated in the age of AI, as ways to cheat the process proliferate. It’s a particularly tricky problem in software development, where it’s now child’s play to ace coding exams with AI, and there are even startups offering those services.
While some companies look for ways to catch the cheaters, financial services company Brex is taking a different approach: It requires that interviewees use AI in coding exams.
This came up in a recent conversation with Brex Chief Technology Officer James Reggio. Brex’s AI-enabled interviews aren’t easy. In some ways, they are a bigger challenge than the previous coding tests.
Candidates are asked to build an entire product in a short period of time, demonstrating both how they can use AI to speed up the coding process and how they think through complex problems. It still requires technical knowledge. AI-generated code isn’t perfect and will almost always make glaring errors in longer projects that need to be corrected.
This method helps the company select candidates more prepared to use AI in their workflows. It’s not easy for trained computer scientists to make that leap, especially because the technology is still so imperfect.
We could see this strategy catch on in other industries and in school exams. For instance, interviews for business consultants might require them to put together a massive research report in a 30-minute interview. Take-home essays could turn into entire books.
Rather than giving people an easy way to cheat, AI could instead just raise the bar.