According to Dongcha Beating Monitor, Bloomberg reports that the generative AI giant Anthropic is implementing a highly controversial and challenging interview process, strictly prohibiting candidates from using AI tools for cheating and introducing a so-called "therapy-like" cultural interview.
With the workforce rapidly expanding in the past six months, CEO Dario Amodei revealed that he spends nearly 1/3 to 40% of his time on cultural development. This cultural interview process carries a veto power and aims to identify candidates highly aligned with the security mission and possessing independent thinking skills.
Candidates must pass through up to 5 rounds of rigorous screening, including signing a Non-Disclosure Agreement (NDA). During the interview, apart from explicit authorization, the use of large language models for assistance is strictly prohibited. In this high-elimination-rate cultural interview, interviewers adopt a rapid-fire questioning style to extract evaluation signals, even interrupting directly when no incremental information is provided. Some candidates revealed that interviewers show no facial expressions or feedback when asking questions, creating significant psychological pressure on the applicants.
The most crucial hurdle in the interview is to assess candidates' values and their awareness of the existential threat of AI. Interviewers present classic professional ethics dilemmas, probing candidates about their emotional responses in moments of conflict, their responses and reflections, aiming to evaluate the candidates' genuine ethical struggles. Moreover, candidates are required to answer questions such as "What unusual beliefs have you held and how did you defend them in difficult situations?" President Daniela Amodei pointed out that the assessment does not seek specific beliefs but evaluates whether candidates can engage in rational debates and focuses on discovering applicants who maintain a healthy skepticism towards the company.
