Thread by @KHS9NE

On new projects, I threw down the ORM and started using plain SQL for a while now.

Lately, Copilot and ChatGPT began to help me to the point I've started writing comments instead of queries and let the AI do the rest, with a bit of polishing later.

And regarding code, we can say the same.

Those 50 bucks a month gave me back hours of "labor work", allowing me to focus on more high-level stuff.

What amazing times.

To not underestimate: my codebases are becoming way more readable now, because I'm forced to write clear docstrings and comments if I want the AI to do the right job quickly.

That was something I've been a bit lacking on 😅

I may be late as a lot of people is already into Copilot, but it took me a while to (accept?) and understand the real potential of the current implementations.

Here, I literally wrote just the method name and yes, the code is good, I can just hit the tab and move further.

The shown method is a 1 to 10 minutes job that, may depend on many factors.

The effort reduction for the human here is about, what, 99%?

How valuable is your time?

I'm stunned.

Later, some performance profiling must be added to the workflow on the daily basis, for the human to evaluate pain points and stuff that needs refactoring.

Here ChatGPT comes in handy, as copypasting inefficient methods to turn into efficient ones is quite a popular feature 🙃

* these tweets have been written down by a human, instructed over the last years by Duolingo AI, and corrected in real-time by Grammarly AI.

#AI #takeover