How AI boosted my software development productivity

For the past few weeks I’ve been using AI as my personal assistant for writing code in my day-to-day software developer job…and boy, I can’t see myself writing code without it anymore.

I know, this sounds a bit extreme, but bare with me while I tell you this short story of how I switched from the idea of AI being something 10 years away into the future to “Is my VS Code alive?”.

I kept a close skeptical eye on this project ever since GitHub announced it on 29th of June 2021, and it intrigued me from the very beginning. My curiosity was finally satisfied last month, when they finally made it available on a subscription-based service for individual developers.

It works as a simple editor extension, you install it, and you start pressing Tabs for autocompletion.

There is no doubt that the AI works best when you need to write some straight forward piece of code, like a sorting algorithm:

And it gets hilarious when he starts testing the code that he wrote, at this point It already starts feeling like cheating.

Of course, recommending some widely used algorithms isn’t really a challenge worthy for any AI, so I went on and created a task scheduling microservice written in typescript with an OOP approach rather than a functional one. Why OOP? Because I was more interested in how he describes things rather than actions.

It was really dumb for a maximum of 15 minutes. The AI keeps track of your project, takes into account your coding style & patterns, tries to understand what you are building and as soon as it does, it feels like you are no longer the only one working.

In my case, it felt that it clicked after defining my database connection and a couple of queries. I was just defining imports, class names, and it already made a good skeleton for what I needed.

It is a double-edged sword as if you are just starting your software development journey you might make substantial mistakes, bad naming conventions, antipatterns, inconsistent usage of different programming paradigms…and the AI will basically do what his master preaches.

But in the same time, if you jump straight into a project, and you are not really aware of the domain, then it’s a marvelous tool.

Remember the simple selection sort use-case and the unit test that he created?
This is the moment when the dilemma starts. How sure can we be sure that he is not biased? What if the test fails? Maybe if the test fails we’ll stop using him, so why would he write a test that fails for a piece of code that he wrote? That wouldn’t be in his best interest, right?

To be honest, I highly doubt it, but, WHAT IF? :)

Related posts:

All items included in Vogue are independently selected by our editors. Nevertheless, when you get something via our retail web links, we might make an affiliate compensation. Few points are rather as…

This question has been asked by a new evil graduate who doesn’t want to pay and I can’t seem to find the answer on google for him. So if a newly graduate happen to have a good credit rating and a big…