How to Embrace ChatGPT in Technical Interviews
Like all of you – we have been experimenting with ChatGPT.
As you may have heard from me before, we at CoderPad aren’t afraid of AI and new technologies. These Large Language Models (LLMs) will not replace developers – but rather make them more efficient. And given that, I’m strongly urging companies to embrace these tools to make technical interviewing better.
If you are asking, “Why would I use ChatGPT in the developer interview process when ChatGPT can pass every technical interview?” The reality is that it’s not entirely true.
ChatGPT is “smart” (shall we say) but not always right. Just like Google search will sometimes return website results that don’t reflect what we were actually searching for. Or code snippets on StackOverflow don’t always meet the need. Or the autocomplete being used to write this article doesn’t anticipate accurately what I want to say. The technology helps but ultimately isn’t able to think or create as critically and with the nuance that the human brain can.
So how do you embrace ChatGPT or other LLMs in interviewing?
- Have a candidate ask ChatGPT to solve a question, get the result, and have the candidate critique the results. How could she or he improve the code ChatGPT provided? This will help you assess the candidate’s skills while also giving you insight into how she or he will fare in the world with additional tooling.
- Have the candidate ask ChatGPT to solve a question and then have the candidate write a readme on how the code works. And the interviewer can grade that answer. This will also give you insight into the candidate’s technical skills as well as ability to document and explain concepts to others.
- Use ChatGPT to help your interviewers refresh or bring new questions. This helps you quickly update questions and have engaging and thoughtful interviews. This also gives interviewers the time to think critically about what concept to ask rather than worrying about the details of the question to test the concept.
You can see in the video below how we experimented with ChatGPT to create a database that can be used in future interviews. (This is way easier and more efficient than having to create a database from scratch or modify existing data to be masqued). Here you have a quick set of data off which to test various dev concepts.
It’s time to stop fearing ChatGPT and assuming that technical interviews are destroyed because of its capabilities. I challenge you to consider embracing ChatGPT to make hiring technical talent faster, fairer, and more real.