AI is the reason interviews are harder now
AI tools are reshaping software engineering hiring by enabling candidates to cheat on remote interviews, mass‑spam applications, and lean on LLMs for algorithmic questions, prompting some companies to respond with harder or more convoluted screening. Commenters argue over whether this actually improves hiring outcomes, with many criticizing LeetCode-style puzzles as already misaligned with real work and suggesting alternatives like in‑person interviews, realistic tasks, and code reviews that test how candidates think and collaborate. Underneath is a broader anxiety: if AI can handle much of the “mundane” coding, employers may raise the bar for human hires and further marginalize those without strong networks or elite credentials.
Are Interviews Actually Harder Now?
- Some argue interviews were already “broken” and AI doesn’t meaningfully worsen things; hiring remains largely about luck or connections.
- Others say AI-enabled cheating forces companies to tighten processes (harder questions, in‑person rounds, more heuristics like school pedigree), indirectly making interviews harder.
- A few experienced interviewers claim interviewing is not harder than past cycles; it’s always been difficult to do well.
AI as Cheating vs Legitimate Tool
- One camp: candidates should exploit AI (LLMs, bots, mass applications) to maximize outcomes; the system is a tragedy of the commons, so rational actors game it.
- Opposing camp: this behavior is abusive, unethical, and degrades the hiring ecosystem; some mention potential blacklisting risks.
- Others take a middle view: using search/AI openly is fine; the core issue is dishonesty and misrepresentation of one’s actual ability.
What to Test When AI Exists
- Some say if AI can solve an interview problem, the interview is flawed; we should design tasks AI can’t trivially solve or that require human judgment.
- Suggested adaptations:
- Code-review exercises (real or synthetic PRs).
- Debugging/bug-hunting tasks on existing code.
- Mixed sets of AI-generated correct and incorrect code, asking candidates to discriminate.
- Assessing how people use their own tools (IDE, Copilot, Stack Overflow).
- Counterpoint: allowing AI in interviews adds noise and obscures individual problem‑solving ability.
Remote vs In‑Person Interviews
- Remote interviewing is seen as easier to game (hidden helpers, LLMs, phones).
- Some advocate returning to in‑person interviews with offline machines and controlled tool access.
- Others note cheating predated AI and even in‑person formats aren’t foolproof.
Broader Hiring Dynamics
- Networking remains disproportionately powerful; this disadvantages immigrants and those without connections.
- Many criticize FAANG‑style LeetCode/hard algorithm interviews as disconnected from real work and serving mainly as high‑pressure filters.
- There’s concern that AI raises the minimum bar: if LLMs can outperform very weak devs, some existing roles may be harder to justify.