Ask HN: What is interviewing like now with everyone using AI?

Hacker News Discussion ↗

Job market and process quality

  • Many describe the market as tight, wage‑suppressing, and employer‑tilted: fewer responses, more “ghost jobs,” frozen requisitions, and 7+ stage processes.
  • Candidates feel more disrespected: slow or no feedback, canned rejections, heavy unpaid take‑homes, and long Leetcode gauntlets for relatively ordinary roles.
  • Common advice: if you’re employed and not being mistreated, “weather the storm” rather than re‑enter the current process.

AI, cheating, and collapsing trust

  • Widespread use of LLMs for resumes, cover letters, rejection replies, take‑home tasks and even live interviews (speech‑to‑text → LLM → teleprompter).
  • Interviewers report obvious tells: delays before answers, eyes tracking another screen, textbook‑perfect but shallow responses, inability to handle follow‑ups or small changes.
  • Some companies are moving back to in‑person or coworking‑space interviews, whole‑screen sharing, and camera‑on policies to limit undetectable assistance.
  • Others argue tools will soon be good enough that a candidate can act as a “ventriloquist dummy,” making detection essentially impossible.

Diverging philosophies on AI in interviews

  • One camp bans AI during interviews to see baseline reasoning, debugging, and coding skills; they treat covert AI use as dishonesty or even “fraud.”
  • Another camp explicitly allows or expects AI use and evaluates:
    • How candidates prompt, constrain, and iterate.
    • Whether they can spot hallucinations, integrate results, and adapt code live.
  • Some design “AI traps” (ethically blocked topics, deliberately underspecified tasks, or interviewer‑supplied LLM interfaces that inject subtle errors) to distinguish understanding from copying.

Backlash against Leetcode and rote tests

  • Many argue Leetcode‑style questions were already misaligned with real work and are now trivially solvable by AI, making them nearly pointless.
  • Some celebrate automated Leetcode solvers as a forcing function to kill that interview style.
  • Alternatives proposed: real debugging sessions, PR/code reviews, extending a small existing app, system‑design tradeoff discussions, and deep dives into past projects.

Portfolios, networks, and bias

  • Some employers now largely ignore resumes and rely on public portfolios (GitHub, blogs, talks) plus internal referrals; if nothing public exists, you may never be seen.
  • Pushback: experienced devs often can’t open‑source their work, lack time or energy for side projects, or choose not to expose personal GitHub; GitHub‑centric hiring is seen as exclusionary.
  • With AI‑distorted signals, referrals, internal applicants, and personal networks appear to matter even more.