--- title: "Interviewing in the AI Era: A Two-Week Plan" url: https://www.thinnd.site/en/journeys/ai-era-interview-two-week-plan/ language: en author: "Nhu Duy Thin (Thinnd)" published: 2026-10-29 updated: 2026-10-29 section: "Journeys" tags: ["interview", "hiring", "AI", "junior", "roadmap", "preparation"] summary: "There is no shared rulebook for coding interviews in the AI era: according to summaries of CoderPad's 2026 report, about 46% of hiring leaders allow AI (broadly or with constraints), 20% decide case by case and 34% ban it, and Google is reported to be piloting a code reading and debugging round with an AI assistant. So the first step is to ask the employer the rules. The two-week plan: ask the rules, practice fundamentals without AI, practice reading and debugging unfamiliar code, practice working with AI in a disciplined way (state a plan, ask small, read, test, explain), mock interviews with a person, and prepare three stories about how you use AI." license: "Cite with the author name and the canonical URL above." --- # Interviewing in the AI Era: A Two-Week Plan > For people about to take a coding interview, especially freshers and juniors. Every technical term gets a short explanation. > > The information about interview policy comes from **summaries and secondary news reports**, not official company announcements. Policies change fast and differ by employer, so **ask your own employer**. The two-week plan is **my suggestion**, not verified on many people. --- ## Quick summary - **There is no shared rulebook.** Some places allow AI in interviews, some ban it, some decide by round. - So the first job is to **ask the employer the rules**. - Whatever the rules, what gets tested most is still: **can you explain code, debug, and verify what AI gives you**. - A **14-day** plan: ask the rules, fundamentals without AI, reading and debugging, disciplined work with AI, mock interviews, prepare stories. - Have an answer ready for: **"How do you use AI?"** --- ## Hard words, plain meaning | Term | Plain meaning | |---|---| | **Live coding round** | You write or fix code during the interview, with someone watching. | | **Code comprehension round** | You are given existing code and must read, explain, debug or optimize it. | | **AI-off mode** | A test platform turns off autocomplete suggestions and the AI assistant. | | **Debugging** | Working out why a program does the wrong thing and fixing it. | | **Complexity** | A measure of how much slower a program gets as the data grows. | --- ## 1. The context: the rules are changing and not uniform Two groups of information, both secondary and to be read with care. - **General statistics.** According to blogs summarizing CoderPad's 2026 report, about **46%** of surveyed hiring leaders allow AI in interviews or assessments (broadly or with constraints), **20%** decide case by case and **34%** ban it entirely. I have not read the original report. - **One concrete example (Google).** According to a report based on an internal document read by Business Insider (not an official announcement), Google is piloting an interview format that lets candidates use an approved AI assistant in a **code comprehension** round. Candidates read, debug and optimize an existing codebase instead of writing from scratch, and interviewers assess how they prompt the AI, the quality of the output and their debugging skills. The pilot targets junior to mid-level roles on select teams in the US. The same coverage mentions Canva and Cognition also allow AI. Two takeaways, both to be understood carefully: 1. **There is no shared standard.** Your colleague tells one story; the company you applied to may do something else. 2. **Where AI is allowed, the grading is still on very familiar things:** reading, debugging, explaining, verifying. That fits other data: the Stack Overflow 2025 survey found 46% of users do not trust AI's accuracy, so interviewers want to see how you **verify**. Also, per the summaries, test platforms have AI-off modes and many live rounds still use them. That is another reason not to skip practicing fundamentals. --- ## 2. The 14-day plan Assume about **1 to 2 hours a day**. If the interview is further off, stretch the plan. | Day | Work | Output | |---|---|---| | **1** | Ask the employer the rules (section 3) | A list of rules and rounds | | **2 to 4** | Fundamentals **without AI**: write small functions yourself, explain complexity, write tests | Three problems solved and explained aloud | | **5 to 7** | **Read and debug unfamiliar code**: each day a 50 to 150 line piece by someone else | Three debugging sessions with notes on the steps | | **8 to 10** | **Work with AI in a disciplined way** using the six-step routine (section 4) | Three problems done with a log | | **11 to 12** | **Mock interviews** with a person, one with AI and one without | Written feedback | | **13** | Prepare **three stories** (section 5) | Three short answers | | **14** | Rest, technical check (network, environment), sleep early | Ready | --- ## 3. Day 1: six questions for the employer Email or ask the recruiter, politely and specifically: 1. What are the interview rounds and how long is each? 2. **Which round allows AI** and which does not? 3. If allowed, **which tool** (provided by the company or your own)? 4. Is there **screen sharing**? Is the conversation with the AI recorded? 5. A comprehension round or writing from scratch? In which language? 6. What do interviewers look for in a candidate at my level? Asking this **does not cost you points**. It shows you are serious. If there is no clear answer, prepare for both cases. --- ## 4. Days 8 to 10: working with AI in a disciplined way If AI is allowed, **how you work** will be visible. A six-step routine (my suggestion): 1. **Restate the problem** in your own words, and ask about anything unclear. 2. **State your plan before asking AI.** Two or three sentences on how you intend to proceed. 3. **Ask small.** Ask AI for a specific part, not "do it all". 4. **Read the result** like a grader. Say aloud what you see and what you doubt. 5. **Test.** Run examples, try odd cases (empty, very large, wrong type). 6. **Explain** why it is correct, and what risks remain. The list of [15 checks for reviewing AI-written code](https://www.thinnd.site/en/engineering/reviewing-ai-written-code/) works as a reference for steps 4 and 5. **Practice both ways.** Do one problem with AI and one without. The point is not which is faster but **whether you can explain both**. An Anthropic trial (January 2026) found that people who let AI write for them scored lower on a knowledge quiz than people who wrote it themselves, especially on debugging questions (a small sample, and I read only the summary). That is why you should practice writing it yourself first. --- ## 5. Day 13: three stories about AI Interviewers often ask "how do you use AI". Prepare three short pieces: 1. **How you use AI.** Three parts: what for, how you verify, when you do not use it. - *Example (illustration):* "I use AI to explain unfamiliar code and to write sample tests. I always run the tests and reread every line before using it. I do not use it when I am learning a new concept, because I want to try it myself first." 2. **A time AI got it wrong and you caught it.** Be concrete: where it was wrong, how you noticed, what you did next. This is often **the strongest signal**, because it shows you really verify. If you have no real story, create one by practicing days 8 to 10 and logging it. 3. **A decision you made yourself** (a trade-off, what to drop). To show the part AI did not do for you. **Do not invent stories.** If you have none, say honestly that you are just starting and how you have practiced. Interviewers often spot a made-up story. --- ## 6. A self-score after each practice I borrow the **pass or fail** grading (as in the post on [writing evals for an AI feature](https://www.thinnd.site/en/engineering/writing-evals-for-ai-features-thirty-real-cases/)) to avoid scoring by feel: | Question | Pass? | |---|---| | Did I restate the problem and ask about what was unclear? | | | Did I state my plan **before** asking AI? | | | Did I read every line of the result? | | | Did I try at least one odd case? | | | Can I explain why it is correct, aloud, without looking at the code? | | Two sessions in a row with a failing item: practice that item more next time. --- ## 7. Easy mistakes - **Using AI without saying anything.** Silence stops the interviewer seeing what you think. Speak aloud. - **Accepting the first result.** This is the easiest thing to catch. - **Practicing only with AI.** Many live rounds still turn AI off. - **Not asking the rules.** Guessing the rules wrong costs points for no reason. - **Memorizing answers.** Interviewers often ask "what if the requirements change". - **Inventing AI stories.** Better to say it honestly, even if small. --- ## 8. A balanced view - **The policy information is secondary.** I have not read CoderPad's original report, and the news on Google rests on an internal document read by the press. The numbers may have changed by the time you read this. - **Every company differs.** Some rounds still turn AI off, others ask directly how you use AI, so do not prepare for only one kind. - **Two weeks is a suggestion.** If your fundamentals are solid you can shorten it. If not, look at the [12-month plan](https://www.thinnd.site/en/journeys/fresher-ai-era-12-month-plan/) first. - **The plan is untested.** It rests on reasoning from news reports and small studies, with no data on who gets hired thanks to it. - **Interviews have many non-technical factors:** communication, attitude, fit. This plan covers only the technical and AI part. --- ## 9. Conclusion The interview rules are changing, but what is tested is fairly familiar: **do you understand, can you debug, can you verify**. AI just makes the last question clearer. Ask the rules, practice fundamentals, practice working with AI in a disciplined way, and prepare three true stories. Start today: write the six questions in section 3 and send them to your employer. --- ## References - [Google's AI-Assisted Coding Interview (2026 Guide), a roundup](https://careers.northeastern.edu/blog/2026/05/13/googles-ai-assisted-coding-interview-2026-guide/) - [AI Tool Policy Tracker: Which Tech Companies Allow AI in Interviews (techinterview.org, secondary, unverified)](https://www.techinterview.org/post/3233474911/ai-tool-policy-tracker-tech-companies-2026/) - [Stack Overflow Developer Survey 2025, AI section](https://survey.stackoverflow.co/2025/ai) - [Anthropic: AI assistance and coding skills (January 2026)](https://www.anthropic.com/research/AI-assistance-coding-skills) - Related posts in this blog: [Fresher in the AI era: a 12-month plan to become a junior](https://www.thinnd.site/en/journeys/fresher-ai-era-12-month-plan/), [Reviewing AI-written code](https://www.thinnd.site/en/engineering/reviewing-ai-written-code/), [A fresher and a senior talk about learning in the AI era](https://www.thinnd.site/en/conversations/fresher-senior-dialogue-learning-in-ai-era/)