What is GitHub Copilot and why should I care?
Because its so cool 🤖 + 👾 = 🙌
I am lucky enough to be able to play around with the technical preview of GitHub Copilot.
Since most of the time I am coding as a hobby because I do more architecture work currently.
This is like having a buddy right there next to me or maybe a really smart buddy 🙉
So how’s it work, well I type this in as a comment right..
// check valid email address
hit enter for new line and Copilot picks up what you are trying to do and gives you options you can choose from. It’s powered by OpenAI Codex, which is trained on billions of lines of public code, so it understands context and can suggest whole functions, not just autocomplete.
The suggestions appear in a dropdown on the right and provide me a list of what’s possible based on the natural language comment I wrote.
I mean it’s scaffolding the entire function for me, WOW!
I scroll through on the right hand side looking at possible solutions and accept a solution, hehe the first one since I’m lazy.
Now isn’t that just magic 🌈
I mean takes the whole copy and paste from stackoverflow to a whole new level don’t you think?
How does it work?
Well the website below has all the details so check it out below but this diagram says it all. GitHub Copilot uses machine learning models trained on natural language and source code from publicly available sources, including code in public repositories on GitHub. The AI model is based on OpenAI Codex, which has been trained on terabytes of public code.
The process works like this:
- Context Analysis: Copilot analyzes the code you’re writing, including comments, function names, and the surrounding context
- Pattern Recognition: It identifies patterns from its training data that match what you’re trying to accomplish
- Code Generation: Based on the patterns, it generates code suggestions that fit your context
- Ranking: Multiple suggestions are ranked by probability and relevance to show you the best options
Conclusion 🐒
Conclusion Monkey says it’s just like having a friend living in the IDE to help me when the days have been long and my brain is tired but I want to get coding done, just so awesome!
Now, there are some important considerations to keep in mind. Data privacy and security are crucial topics when using AI-powered tools like this. GitHub Copilot processes your code to provide suggestions, but according to their privacy policy, they don’t store or share your code. However, the suggestions themselves are generated from patterns learned from public repositories, so there’s always a chance you might get suggestions that resemble existing code.
Also worth noting - you should always review the suggested code before accepting it. While Copilot is incredibly smart, it doesn’t understand your specific business logic, security requirements, or the broader context of your application. Think of it as a very sophisticated autocomplete, not a replacement for good engineering practices.
Data privacy and some other security stuff I’ll dive into deeper later but there are tons of people on the net smarter than I am diving much deeper on this topic, just google-bing it, if you are interested or check out the below links.