Curious about "no-AI" policy. From my experience interviewing lately AI is allowed,but the task is usually big in a small time window. So you quickly spin up the project but I see most of work in QA
A friend at a nearly-FAANG said not using AI tooling in the interview is now an automatic fail. “Not that you’d be able to complete the task on time without it anyway”
That's a good policy. If that's what they do in the job later as well, then it allows LLM-skeptical applicants to immediately move on.
An interview is a two way street. I'd like the company to present itself authentically so that I don't waste time if they eventually turn out to have a culture I don't like, e.g. demanding LLM coding assistent use.
How does that work? I assume it's not Leetcode anymore then? Current LLMs mostly one-shot these types of algorithmic exercises, except maybe for the most difficult ones.
I interviewed at a place that proudly stated that they have a goal that x% new LOC are LLM-generated. They didn't say what x was, but implied it was high.
For general software engineering, the accepted thought is you're not really an engineer without the use of AI. Because engineering consists of using the right tool for the job, applying best practice, making tradeoffs, and justifying every decision from a technical standpoint. And it's anti-engineering to write the code yourself rather than putting your ego aside and taking advantage of AI's huge productivity gains.
Game dev is... different. Game devs fancy themselves more as artists, and using generative AI is an affront to those sensibilities.
We’re using AI tools heavily but also ask candidates to code without it in interviews, and I think it’s reasonable.
Even if you are primarily using codegen, your own coding ability, taste, problem solving, etc. are still deciding inputs to the quality of the final result. And it’s much easier to assess these things in an hour when a human is writing and debugging a relatively small amount of code.
AI tools just produce too much code in a short time. It’s hard to assess what the candidate’s quality bar and attention to detail are really like when there’s so much code to wade through. Anyone can vibe code, but not many can do it without creating mountains of tech debt… the ones who can are usually good programmers with or without AI.
Does it also mean no-Googling? Because the first result is usually LLM-generated. Should you skip to the next page and read only Experts Exchange pages?
testing22321|3 months ago
teiferer|3 months ago
An interview is a two way street. I'd like the company to present itself authentically so that I don't waste time if they eventually turn out to have a culture I don't like, e.g. demanding LLM coding assistent use.
qsort|3 months ago
Vegenoid|3 months ago
bitwize|3 months ago
Game dev is... different. Game devs fancy themselves more as artists, and using generative AI is an affront to those sensibilities.
danenania|3 months ago
Even if you are primarily using codegen, your own coding ability, taste, problem solving, etc. are still deciding inputs to the quality of the final result. And it’s much easier to assess these things in an hour when a human is writing and debugging a relatively small amount of code.
AI tools just produce too much code in a short time. It’s hard to assess what the candidate’s quality bar and attention to detail are really like when there’s so much code to wade through. Anyone can vibe code, but not many can do it without creating mountains of tech debt… the ones who can are usually good programmers with or without AI.
zerr|3 months ago