Overview
ChatGPT, Github CoPilot, and other AI-powered coding assistants have taken the world by storm. This survey reveals how software developers are embracing these tools to boost their productivity and learn new technical skills.
Here are the most eye-opening findings from our report:
Report sections
How developers use AI coding assistants
81% of developers surveyed say they use AI-powered coding assistants. That’s nearly twice the percentage of developers who said they used AI tools in Stack Overflow’s 2023 Developer Survey, conducted 6 months earlier.
Here's how they use them:
Learn new technical skills / knowledge
Generate boilerplate code
Practice new technical skills / knowledge
Debug / check code for errors
Explain, comment, and/or document the code
Rewrite code using correct style / optimal code
Generate regular expression (regex)
Write test cases
Implement algorithmic logic
Rewrite code in different languages (code translation)
Write code to use an existing third-party API
Check vulnerabilities
Refactor legacy code
Generate service implementation code based on an API specification
Other (2%)
Key takeaways: The most common use of AI coding assistant tools isn’t actually for writing, commenting, or debugging code: it’s for learning. Over three-quarters of developers who use AI-powered coding assistants use them to learn new skills or knowledge. This stands in contrast to early studies of these tools from Stack Overflow, GitHub, and Google, which focus on the benefits for developers’ productivity.
Top AI coding assistant tools
Among AI coding assistant tools on the market today, one clearly dominates: ChatGPT.
These are the top 5 tools developers use:
ChatGPT by OpenAI
Github Copilot
Bard by Google
Visual Studio Intellicode by Microsoft
Claude by Anthropic
Adoption and frequency of AI coding assistant use
Among developers who say they use an AI coding assistant:
49% use them daily —and another 39% use them weekly
58% started using these tools within the last 6 months
71% use these tools more often today than when they first started
What developers are excited about for the future of AI
We asked developers: What are you excited about when it comes to AI uses for software engineering?
Here’s what they said:
Boosting my productivity while coding
Personalized tutoring / teaching
Boosting my productivity, for non-coding tasks
Improving the quality of my code
Writing code
Opportunities for practice-based learning
Helping with parts of my job or school that I don’t enjoy
Key takeaways: As in other studies of AI tools, developers are excited about the potential for greater productivity while coding.
A senior software engineer who uses GitHub Copilot describes the experience this way:
“[With Copilot] I have to think less, and when I have to think it’s the fun stuff. It sets off a little spark that makes coding more fun and more efficient.”
What makes developers nervous about AI
Here’s what they said:
I have concerns about quality of AI-generated code
I have concerns around security and/or privacy
It may result in lower demand for hiring software engineers / developers
It means I’ll need to learn new skills to deliver value to my employer or find a job
I have other ethical concerns
I'm not nervous or concerned
Key takeaways: The biggest concern of developers surveyed was the quality of AI-generated code This aligns with Stack Overflow’s finding, where only 2.85% of developers surveyed “highly trusted” the accuracy of AI output.
Conclusion
Developers today are rapidly adopting AI-powered coding assistant tools like ChatGPT in their daily work. Like other early studies of these tools found, this survey finds that devs are using these tools to boost their productivity. Most commonly, however, devs are also using AI coding assistant tools to learn and practice.
This new finding has major implications for how developers level up their technical skills and learn to code. The widespread use of tools like ChatGPT for learning indicates a shift away from traditional lecture-based instruction—and toward a future where learning is highly personalized, interactive, and practice-based, all facilitated by AI tools.
Methodology
This survey was conducted by CodeSignal between November 20 and December 4, 2023, via the Qualtrics platform. The survey was emailed to a sample of individuals (n=30,837) who had taken the General Coding Framework (GCF) or Industry Coding Framework (ICF) assessment on CodeSignal, or who had used CodeSignal Arcade within the past year. Respondents were offered an incentive of entry into a raffle for one of 5 $50 Amazon gift cards. The survey was completed by 1,021 respondents.