What is the general consensus / what do you feel a...
# random
g
What is the general consensus / what do you feel about candidates using GitHub Copilot to solve take-home interview tasks and assignments?
l
If you can't check / enforce stuff on "take-home" tests, does it matter? 🙂 In my opinion, candidates should use whatever tools they would use in their normal day-to-day operation. Doesn't matter to me personally.
âž• 7
q
We prefer having a live coding round over giving take-home assignments to the candidates. Having said that, even if the candidate uses GitHub Copilot or for that matter anything else, the way to draw that out is to thoroughly evaluate them based on what they have done. That will separate chaff from wheat.
âž• 1
g
What would your reaction be if you have a candidate using Copilot in a live coding session?
q
Interesting question. We do live coding rounds using Jupyter notebooks (I am usure if you can use Copilot in the notebook). However, assuming that you can use, here is what I would watch out for:- 1. If the candidate is using Copilot to speed up mundane task but does the necessary heavy lifting where it is required. I would be Okay with it because it shows that the candidate wants to have control over items that are crucial for successful development, while automating the boring stuff. 2. If the candidate ends up using Copilot for everything then that's a candidate I would not proceed forward with because it shows that the candidate is relying heavily on a third party tool rather then owning the development process. During our live coding rounds, we allow candidates to use google for referring documentation wherever needed. However, you might be surprised to know that quite a few of them don't know how to look for what they want to do using the documentation.
âž• 1
e
I am okay with any and every tool which developer use in his day to day job. I don't ask developer to use any specific tool in live coding session, he can use his normal editor to anything where he can comfortably write and execute the code for given problem.
âž• 2
h
We used a 2 stage interview. The takehome problem could be solved by the candidate for first pass. Once it's reviewed, we have a live round where we extended the original problem statement and have them proceed from where they left off.
âž• 1
This method let's us see how their actual writing process looks like, without unnecessarily restricting the candidates toolset
âž• 1
r
'