Summary of Key Points
Anthropic recently raised $65 billion in funding, valuing itself at $965 billion, surpassing OpenAI to become the most valuable AI startup in the world. Its recruitment standards are exceptionally high: AI is completely banned during the entire interview process, with a focus on using "cultural interviews" to identify candidates whose values align with the company's and who possess independent thinking skills. This unconventional approach has helped Anthropic retain a large number of top talents (with an 80% two-year retention rate, the highest among its peers), and it has even attracted notable individuals such as OpenAI co-founder Dario Karpathy. Unlike Google, which allows the use of AI to assist in interviews, Anthropics believes that what is most scarce in the AI era is not "people who know how to use AI," but rather "people who can think for themselves even when AI is turned off."
Detailed Analysis
1. No AI in Interviews? Anthropic Fears "Outsourced Thinking"
Anthropy prohibits the use of AI throughout all five rounds of interviews for a simple reason: it wants to see your true abilities and thoughts, not just what AI can make you appear to be capable of. For example, technical interviews do not use AI to directly assess your problem-solving skills; cultural interviews avoid AI-generated "politically correct" responses. After all, while AI can produce well-written articles from any perspective, it cannot convey your true beliefs—something that Anthropics values greatly.
2. Cultural Interviews Like Psychotherapy: Screening for Like-Minded Values
Cultural interviews are a critical part of Anthropics' recruitment process:
- No Technical Skills, Only Core Beliefs: They do not ask about the quality of your code but rather questions like, "What unusual beliefs do you have?" and "Have you ever persisted in the face of discomfort?" or "How would you make a decision in an ethical dilemma?" For instance, candidates might be asked whether they would hide or disclose a user data breach. The interviewers are looking for genuine deliberation and principles, not just the correct answer.
- Single-Vote Disqualification: Even if you pass all technical interviews, a poor score on the cultural round can lead to rejection. Interviewers may come from marketing departments (not technical roles) because the company believes that shared values are more important than technical skills. Anthropics views AI safety as its core mission, not just a means of making money.
- Encouraging Questioning: Interviewers will deliberately point out flaws in Anthropics' practices, and well-reasoned criticism is seen as a positive sign of independent thinking.
3. The Effect of This Recruitment Method
The data speaks for itself:
- High Employee Retention Rate: With an 80% two-year retention rate, Anthropics has the highest rate in the AI industry (whereas peers typically have rates below 60%).
- Strong Talent Attraction: The number of employees who have switched from OpenAI to Anthropic is eight times higher, and from DeepMind, it is eleven times higher. For example, Workday's CTO gave up a senior position to become a regular engineer at Anthropics; Karpathy joined from OpenAI—these individuals were attracted by Anthropics' values, not just the salary.
4. Going Against Google? Two Companies with Opposite Views on AI Talent
In the realm of AI recruitment, Google and Anthropics have taken completely opposite approaches:
- Google: Allows the use of its own AI, Gemini, to assess "AI proficiency" because daily work involves human-AI collaboration, so interviews need to simulate real-world scenarios.
- Anthropics: Bans the use of AI because it believes that while AI can assist with execution, it cannot make critical judgments for you. Google values people who can write code using AI; Anthropics values those who can think about the major risks associated with AI even without its help.
5. The Most Valuable People in the AI Era
Anthropics' recruitment philosophy addresses a fundamental question: as AI makes tasks like coding and writing more accessible, what becomes scarce? The answer is your independent thinking, strong values, and the ability to identify large-scale risks. For instance, a researcher who only discussed the emotional dependence of users on AI was not hired because Anthropics is concerned about the fundamental risk of AI getting out of control—these are issues that AI cannot predict for you; they require your own critical analysis.
In conclusion, don't just focus on learning how to use AI; instead, cultivate what AI cannot take away: your principles, judgment, and your unique perspective on the world. These are the truly valuable assets in the AI era.