My Experiments with AI
I’ve been experimenting with AI In a scattershot manner. i’m learning not to try to use AI to create a new project from scratch. Instead, I like to bootstrap my new code from working notes, scripts, and projects. Then I use AI to help me add new features to an existing project.
I have lots of scripts and starter projects that provide the framework on which to build a complex project. With those pieces in place, AI becomes useful.
It doesn’t matter whether I use Bing Chat, Google Bard, or GitHub Copilot. They are all good at looking at existing code that works and using that code to help me generate new features. But GitHub CoPilot has the edge because it has the best view of my code. It really helps me get a lot of work done fast.
GitHub Copilot is built on top of the OpenAI Codex and provides enhanced autocompletion inside Visual Studio code. You can trigger it either by starting to write code, or writing comments for the code that you want to create. It excels at certain tasks such as helping you write unit tests. Find more information on by visiting pages like this one by LadyKerr entitled Getting Started with GitHub Copilot Chat in VSCode
Others can go to the Copilot website on GitHub and sign up for a free trial. After the trial, copilot is $10 a month or $100 a year. To learn the status of your account, sign into GitHub and visit the following pages:
To install GitHub copilot in Visual Studio Code, go to the Extensions Marketplace in VS Code (Ctrl-Shift-X) and search on GitHub Copilot.
A good way to get started would be to build a simple project that you know how to complete and use the copilot suggestions to expedite the process. For instance, a teacher could take a simple Getting Started project and show how Copilot makes it easy to create.
** Use Firefox when logging into openai.com.