AI coding assistant

Monday. March 27, 2023 - 1 min

ChatGPT has plenty of compelling use cases, performing well at many of them. Many who work with words are loving using the tool to help with writing. Talking with the developer community, there is serious doubt as to whether it (or more accurately a future version) will “destroy” the software industry in the coming years.

Who knows 🤷‍♀️

But talking to some people, the coding abilities of the Large Language Model (LLM) are dismissed out of hand as nonsense. I’m not sure these people have taken the time to understand how best to use the technology.

I’ve used it to help write tons of words for work in recent months, becoming super comfortable using it to help with sales and marketing-type tasks. Everyone around me seems happy with the output. Yesterday I applied this prompting knowledge to coding for just over an hour.

It was super impressive!

Everything prompted for was in javascript and used native APIs, alongside some supporting HTML & CSS. This allowed the system to sidestep library methods polluting responses. There was also a bit of standard node.js/express to provide a simple backend. Nothing super fancy on either side of the stack mind, but enough to fill an afternoon sat in front of a glowing rectangle.

Bottom line. It worked and took less time than without ChatGPT. Bearing in mind that I’m comfortable coding everything I asked for, if the same process was applied to something outside of this comfort zone, I would have been far more productive.

Suppose ChatGPT and other LLMs improve even slightly over the coming years. In that case, the people working in the software industry who are dismissing it are in for a shock.

If you’ve dismissed ChatGPT’s coding ability, have another look. If you don’t, you might be out of a job in a few years. Then again. You might not.

Isn’t the future fun.