When Interviews Become “Homework Assignments”: The New Workplace Phenomenon Behind AI Chat Logs and How to Overcome It
Hello everyone, I’m your financial journalist and economist friend. Today, we’re going to discuss a topic that’s causing quite a stir in the workplace: employers are now asking candidates to submit their AI chat logs from the past six months.
It sounds like something out of a science fiction story, but it actually happened during this year’s campus recruitment process. As an observer, I find this particularly interesting because it mirrors the deep-seated anxieties in the workplace regarding technology, privacy, fairness, and the concept of “involution” (where everyone is constantly competing with each other).
Don’t let the fancy term “AI” scare you. Let’s strip away the technical jargon and break down what’s really happening, as well as how ordinary people can respond to this trend.
---
Summary of the Main Points: A Workplace Game Triggered by “Laziness”
In simple terms, interviewers are in an awkward position. They want to assess a candidate’s ability to use AI, but they don’t know how to ask the right questions or are afraid that candidates might cheat during the interview. So, companies have come up with a “shortcut” – asking candidates to provide their chat logs with AI over the past six months.
Although it’s labeled as “optional,” in the fierce competition, it’s almost become a mandatory requirement.
The key points of this article are quite insightful:
1. This approach is naive and unreasonable: Chat logs don’t reflect real capabilities; they pose risks of privacy breaches, data fraud, and a lack of clear evaluation criteria.
2. **It’s a classic example of the “Goodhart’s Law”: When an indicator becomes a criterion for evaluation, people tend to focus on making that indicator look better rather than genuinely improving their skills.
3. Candidates can use technology to their advantage: Since interviewers might also use AI to evaluate the logs, candidates can use AI to optimize their own records and thus stand out in terms of “performance.”
4. The right way to interview is through scenario-based testing: The real test should be to have candidates use AI to solve problems in a simulated context, not just review past conversations.
---
Why Submitting Chat Logs Is a Bad Idea?
The “Dashcam” Metaphor: Knowing the Past Doesn’t Mean You Can Drive
The article makes a great analogy: Just because a driver submits their dashcam footage doesn’t mean they can actually drive well.
Similarly, if you’ve talked to AI 1,000 times in the past six months, it might have been about things like “What’s the weather today?” or “Write a poem for me” or “How to stir-fry potatoes.” These logs may seem extensive and indicate diligence, but in a work context, you might not know how to break down a complex business problem, let alone identify when AI is giving misleading information or how to protect sensitive company data.
- In plain terms: It’s like someone who practices driving in a simulator every day, getting their fingers worn out from the constant practice. But when they actually get behind the wheel, they might not even know how to use the gas and brake properly. Frequent use doesn’t equal proficiency. Companies want people who can get the job done, not just those who talk to AI the most.
The “Black Box” of Fairness: Your Tools, My Bias
There are various AI tools out there – ChatGPT, Claude, Gemini, as well as domestic options like Kimi and DeepSeek. Some people pay thousands of dollars per year for advanced memberships, while others use free versions.
- Potential bias: If companies use these logs for evaluation, it can lead to hidden biases, such as the assumption that those who use foreign top-tier models are more capable or that those who use paid versions are more professional. But the key lies in how you use the tools. Someone who uses the free DeepSeek but asks clear, meaningful questions might be much more effective than someone who uses ChatGPT Plus but only asks for summaries.
- Lack of standards: It’s easy for companies to ask for this data, but there’s no uniform way to evaluate it. It’s like an exam where some students use pens, others ballpoint, and others pencil; the teacher can’t give higher scores just because the pen writing looks better; the focus should be on the accuracy of the answers.
The Low Cost of Faking: Easier Than Faking a Resume
Faking a resume requires fabricating experiences and lying. But faking AI chat logs is much simpler:
- Editing: You can delete embarrassing conversations and keep only the ones that seem profound.
- Copying: You can even ask a friend for a “good” chat log and modify it to submit.
- AI Generation: You can use AI to create a conversation history that looks highly professional.
Interviewers don’t have the time to read through all the logs, let alone verify their authenticity. It’s like asking a police officer to investigate a case when the suspect submits a bunch of self-made diaries – would the officer believe them?
The Exposure of Privacy: Your Secrets Are Revealed
Although the article says that privacy can be compromised in many situations, this still poses a significant risk. Your AI chat logs may contain personal emotions, family details, and even complaints about the company or sensitive views on the industry. Sharing these with potential employers is like exposing your private life.
- Risks: If you’ve ever asked about how to resign gracefully or how to argue with a boss, these logs could end your career before it even starts.
---
Why Do Companies Do This? “Laziness” and “Information Overload”
Interviewers’ “Competence Anxiety”
Many interviewers are themselves new to using AI. They recognize its importance but don’t know how to ask effective questions:
- Traditional questions don’t work: Asking “Can you use AI?” is too vague; everyone would say yes.
- Scenario-based questions are difficult: A truly effective interview would be: “Imagine a customer complaint; use AI to draft a response and explain your thought process.” But this requires strong business understanding and skills, which many HR professionals or junior managers lack.
- The Easy Way Out: Since they don’t know how to ask, they ask candidates to submit the logs, thinking it’s a simpler way to assess them.
The Workplace Version of Goodhart’s Law
> Goodhart’s Law: When an indicator becomes a goal, it stops being a good one.
Once “AI usage history” becomes a hiring criterion, candidates start focusing on “performing” rather than actually using AI.
- Result: Companies receive a collection of well-packaged “AI usage scripts” rather than a true reflection of their capabilities.
- Irony: Companies aim to find people who can use AI, but end up hiring those who are best at “ beautifying” their logs. It’s like selecting a long-distance running champion by asking for screenshots of their running apps; everyone just focuses on increasing their step count, not on actually running.
---
How Can Candidates Respond?
Since companies have come up with this unorthodox requirement, candidates can’t fight head-on; they need to use strategy. The article suggests a practical approach:
- Since interviewers will likely use AI to evaluate the logs, let’s use AI to optimize them first.
- Pre-prepare the Logs: Before submitting, select meaningful conversations that demonstrate your ability to analyze, solve problems, and verify data.
- AI Review: Feed your logs to another AI (e.g., use Claude to analyze ChatGPT’s output) and ask it for feedback on what makes you look professional and what needs improvement.
- Optimize Your Narrative: Adjust your questions and reorganize the conversations to better reflect a “professional user’s” style.
- In plain terms: If the judges use AI to evaluate your photos, you should make sure your photos are well-composed and well-lit. Use AI to improve your materials before submission.
Distinguish Between Work-Related and Casual Conversations
- Submit Relevant Content: Only submit records related to the job and your thought process, not trivial conversations about recipes or gossip.
- Mask Sensitive Information: Remove any company secrets or personal details.
Develop a Habit of “Keeping Track”
Interviews are essentially a way to verify your resume. If you regularly use AI at work, develop the habit of saving important conversations.
- Create a Portfolio: Organize your AI-related work – solutions you’ve found, code you’ve written, and texts you’ve generated – into a portfolio.
- Be Ready to Show: When interviewed, you can provide these examples and explain how you used AI to solve problems.
---
Suggestions for Companies and Interviewers: Back to Scenario-Based Interviews
Although the article mainly criticizes companies, it also offers constructive solutions. For companies that truly want to hire the best, asking for logs is a last resort; conducting scenario-based interviews is the way to go.
- Scenario-Based Interviews: Provide candidates with a real, simplified business scenario and ask them to use AI to generate solutions.
- Example: “Imagine we’re promoting a new product for Gen Z. Use your AI to generate three marketing slogans and explain your reasoning and why you chose each one.”
- Evaluation Points: How well you provide context, think critically, and apply the AI’s output to the business.
- Advantages: This method prevents fraud, requires immediate thinking, and tests real capabilities.
- Prove the Reasoning: Even if logs are reviewed, ask follow-up questions:
- “Why did you ask that question of AI?”
- “How did you adjust the AI’s initial answer?”
- “How did you verify the data?”
- “What would you do if the AI’s output involved sensitive information?”
- Respect Privacy: Offer an optional option and clearly state that the data will only be used to evaluate AI skills and will be kept confidential. Provide an alternative option, such as a live demonstration.
---
In Conclusion: Technology Is a Tool, People Are the Core
This whole incident around AI chat logs reflects the fact that technology is evolving faster than management strategies.
Companies want to use AI to improve efficiency but haven’t yet learned how to effectively assess people’s AI skills. Candidates are forced to adapt to this new form of “involution,” where they need to not only be capable but also able to demonstrate their ability to use AI effectively.
For everyone: Don’t panic; asking for logs is just a temporary issue. In the long run, the ability to solve problems on the spot is what truly matters. Focus on developing essential skills like logical thinking, problem-solving, and critical analysis. If you have to submit logs, use technology to your advantage, but the ultimate goal is to show your ability to control AI, not just be recorded by it.
Remember: Interviewers want to see how you think, not how many days you’ve talked to AI. Invest in improving your thinking skills; that’s far more valuable than beautifying your chat logs.