Take-home assignments used to be a great way to evaluate candidates. They gave you insights into a person's coding style and problem-solving process.
In the past, candidates often spent about a week on these tasks. They poured in effort, wrote original code, and delivered thorough solutions.
Now, with AI and Large Language Models, everything has changed. People are completing these assignments in just a day or two. They simply prompt an AI to build the bulk of the application, this makes it impossible to gauge their real skill level.
The bigger issue is that it still leaves many questions unanswered, Hiring managers usually end up doing another interview round. They want to see how the candidate actually thinks, solutions and codes. That extra round is a waste of time for both sides.
Instead, a better approach is pair programming, where you and the candidate work together on a small project. This way, you can see their thought process and coding style in real-time. You set up a small MVP with help from AI before the interview, you and the candidate build a new feature together.
By watching them work live, you get immediate feedback, You can decide if they are a good fit more quickly. You avoid the confusion of AI-assisted take-home tests and you save everyone time in the process.
In today's world, take-home assignments no longer serve their original purpose, It's time to move on to more practical and direct approaches. Pair programming with a ready-made MVP app is a setup that is working out brilliantly for me. It's efficient, transparent, and fair for both parties.
I hope this approach helps you too or you have a better one to share. Do let me know.