Anthropic · Primly Community

Anthropic behavioral interview questions and values, here's what actually came up

nonprofit_nia · 4 replies

I'm not a SWE. I interviewed for an ops/program management role and the behavioral portion was surprisingly substantial, so I wanted to share what actually came up. Most posts here focus on the engineering side.

First: Anthropic's behavioral questions are specifically calibrated around their stated values. Things like intellectual curiosity, willingness to engage seriously with uncertainty, and what they sometimes describe as being willing to hold multiple framings of a problem at once. If you've read any of their public writing on AI safety, you'll notice their values language shows up directly in the interview.

Questions I got (paraphrased): Tell me about a time you had to make a decision with limited or conflicting information. What was your process? Describe a situation where you changed your mind about something important. What caused the update? Tell me about a time when the right thing to do was unclear and there was no obvious correct answer. How did you navigate it? How do you think about working in an area where the stakes feel genuinely high?

Those last two felt distinctive. "How do you think about working where the stakes feel genuinely high" is not a question I'd gotten anywhere else. They seemed to want people who take the mission seriously without being weird about it.

What didn't land: Generic STAR answers that didn't show any actual reasoning or intellectual engagement. I've heard from someone who went through around the same time that very polished, rehearsed answers got lower scores. The interviewers would probe until they hit the real reasoning.

One thing I noticed: Every interviewer I spoke with was able to articulate WHY they worked at Anthropic. Not a canned answer. That was a signal to me about the culture.

Happy to share more about the non-engineering interview track if there are questions.

4 replies

pm_priya

"Willing to hold multiple framings of a problem at once" is a very specific skill and also genuinely hard to evaluate in a 45-minute interview. Did they push on your answers or accept surface-level responses?

sam_recovering

The "stakes feel genuinely high" question is interesting. Did it feel like they were probing for stress resilience or more for mission alignment? Those are different things.

nonprofit_nia

More mission alignment in my read. They didn't seem interested in whether I can handle pressure per se, more in whether I'd thought seriously about what it means to work on frontier AI. I referenced some things I'd read about AI risk and that seemed to be the right register for the conversation.

pivot_pat

This is really helpful for non-eng tracks, thanks for posting. There's basically nothing out there for ops/PM behavioral rounds at Anthropic specifically.