Over the past year, generative AI technologies have brought exciting new solutions to software developers. Software engineers can now take advantage of a variety of tools to help generate code automatically.
Some, like GitHub Copilot, are AI tools designed specifically and solely for helping to write code. Others, such as ChatGPT, are general-purpose AI solutions that can churn out code, but can do many other things as well.
The question for today's developers has become: Which type of AI-assisted tool generation tool is best? Are solutions like Copilot better than ChatGPT for software development because the former were designed for coding in particular? Or does the flexibility of ChatGPT and similar solutions give them an edge over Copilot and the like?
To answer that question, let's compare the main advantages and drawbacks of both Copilot and ChatGPT from a software developer's perspective.
Pros and Cons of Copilot for Software Development
Copilot is an AI-assisted coding tool introduced in late 2021. It's built into GitHub (which is owned by Microsoft) and is powered by OpenAI's GPT language model.
Once integrated into an IDE, Copilot can automatically make completion suggestions for code that developers begin writing. You can also insert comments describing what you want your code to do, and Copilot will generate code to implement them.
Copilot is designed to learn your coding style over time, so the longer you use it, the better it should become at generating code that is similar in structure and syntax to what you'd produce if you wrote it all by hand.
Advantages of Copilot
The main advantages of Copilot include:
- Direct integration with IDEs, which means there is no need to toggle between different tools when using Copilot to help write code.
- The ability to adapt to individual programmers' coding styles.
- Copilot was trained specifically to write code, and the sense among programmers is that Copilot generally does a better job than ChatGPT of producing code that runs with few, if any, manual modifications.
Disadvantages of Copilot
Copilot also has some disadvantages:
- It only supports certain IDEs. It's compatible with most of the popular IDEs, but if you prefer a more obscure IDE, Copilot may not work for you.
- Although you can describe what you want code to do, Copilot doesn't allow you to ask open-ended questions about coding in the way you can with ChatGPT.
- Copilot doesn't do much to help you design application architectures; it mostly just helps to auto-generate relatively short sections of code once you've already decided how you'll implement your application.
Pros and Cons of ChatGPT for Software Development
ChatGPT, of course, is the chatbot released by OpenAI in November 2022. It's powered by the same AI technology as Copilot. But unlike Copilot, ChatGPT is designed to respond to natural language queries of any type and to generate any kind of text — not just computer code.
Advantages of ChatGPT
For software developers, advantages of ChatGPT include:
- The ability to ask open-ended questions. If you are struggling to get started with a development project, ChatGPT can be a good way to obtain suggestions about how to structure your application. It can also churn out some sample code to give you a sense of how you might begin implementing the app.
- Deeper explanations of what code does. Unlike Copilot, ChatGPT can describe in detail how the code it generates works. This is also an advantage if you are just getting started and need a lot of high-level guidance about how to implement and deploy your application.
Disadvantages of ChatGPT
The main disadvantage of ChatGPT for software development purposes is that, on the whole, it's less likely to produce functioning code. Although there are no hard statistics comparing the code quality of Copilot to that of ChatGPT, the general consensus among developers who have worked with both tools is that code written by Copilot is more likely to compile and run properly than code written by ChatGPT.
That's especially true if you're dealing with complex code that needs to integrate into a broader codebase. In that scenario, Copilot's ability to learn your coding style increases its chances of writing code that will work with your other code. In contrast, ChatGPT doesn't have any context about how you code or which other code you have in your repositories, so it can't suggest code that is customized for your needs. Its offerings are much more generic.
Copilot vs. ChatGPT: What to Use When for Coding
In short, Copilot is probably a better solution for developers who already have a coding project underway and need a tool that can help generate high-quality code to complete the project. In contrast, using ChatGPT makes more sense if you're just getting started with a project and want some high-level guidance and sample code.
In fact, perhaps the best way to use Copilot and ChatGPT is to take advantage of both of them. Use ChatGPT to help get your project started, then switch over to Copilot to help complete the more intense coding work.
About the author
Christopher Tozzi is a technology analyst with subject matter expertise in cloud computing, application development, open source software, virtualization, containers and more. He also lectures at a major university in the Albany, New York, area. His book, “For Fun and Profit: A History of the Free and Open Source Software Revolution,” was published by MIT Press.
