Weekly Bytes

A.I. Coding Assistants Are Still Confusing

October 2, 2024

Hello there,

I recently stumbled on 2 articles that painted a clear picture to me that the world is still confused about A.I. and it's potential role as a software engineer replacement. The first article broke down why A.I. coding assistants provide little if any benefit to programmers, particularly in a real-world development setting.

The study referenced measured PR cycle time, or the time that it took to merge code into a repo, and it found no real significant improvements for devs using Copilot.

Probably more interestingly though, the study found a 41% increase in bug rate from developers using A.I. coding assistants. Definitely not an insignificant number.

The second article broke down why A.I. coding assistants do indeed improve developer productivity. The study referenced in this article pointed to a notable 26.08% increase in the number of pull requests completed per week.

However, it also did note that lesser experienced developers tended to get more benefit from A.I. assistants. And from what I read, I could not find mention of error rates either increasing or decreasing.

2 seemingly conflicting points of view, but both somehow making sense. And I think that, really, all they point to is that some companies will benefit from A.I. coding assistants, while others probably won't.

Personally, I don't know if this is even quantifiable. For one, every company out there has their own level of difficulty when it comes to building out their technology stack.

Working on the software for a fully-automated warehouse inventory management system is different than working on an e-commerce platform.

And both have their own unique requirements when it comes to developers and needed skills. One might use plain-old JavaScript, which most language models are vastly trained on due to the languages prevalence around the web. While another company might use a more specialized and lesser known language that A.I. hasn't quite figured out yet.

Productivity is also such a personal and subjective trait to measure, because I could feel very productive today, but in reality my code is terrible, error rates are high and I won't find that out until maybe next week.

All to say, that A.I. probably still has a way to go before it can actually start to steal our programming jobs. We're still safe, for now. ✌️

-- Walt

My latest posts

Check out the latest articles published on the blog:

​

• ​A Look at the Drag and Drop API in JavaScript​
• ​How to Generate GUID's using just JavaScript​
• ​How to Show PDF file Upload Previews​
• ​How to Check for Unused Modules in Node.js​
• ​How to Improve Web Accessibility with Semantic HTML​

​

​

​

Weekly Question

Q: How many programming languages should I learn?

This is one of those questions where often times the person asking wants one of two answers. They either want to hear "All of them. The more the merrier". Or they want a very specific number like "You need to learn 6".

And nobody wants to hear "Just learn 1. But learn it well". That answer is too boring.

But lucky for us, the answer isn't any of those. Because the answer is, it depends on your company and what they're paying you for. If they have a complex system spanning 5 frameworks and 4 programming languages, and they went ahead and hired you for the job, then you're going to be learning 4 languages whether you want to or not.

Most of the time though, you'll probably be working on a single project written in a single language running under a single framework. And that's fine too. I personally worked with C# and ASP.NET exclusively for over 10 years, before shifting over to JavaScript and React.

But in the middle I also dabbled in PHP, Classic ASP, X++ and probably a few more that I've all but forgotten about. And that's not including the C, C++ and Java that I learned in college. But currently, I would not claim to "know" those languages as I have not kept up with their many changes throughout the years.

So maybe I sort of knew 6 or 7 if you tally it up. But more realistically, I know 2. And I know them relatively well.

So don't worry about stacking up languages in case you ever need them, but focus more on the one that is in front of you at any moment in time. That's the one that's probably going to bring in the paycheck.

Have a question? You can submit it here and I might answer it in the next newsletter.

That's all for now. Until we meet again next week 👋