The 2023 Stack Overflow Survey reveals that over three quarters of Professional Developers respondents feel favourable towards using AI tools as part of their development workflow and 44% are already using tools such as ChatGPT and GitHub Copilot. However, only 42% trust the accuracy of the output of these tools!

With its huge community of developers, both professional and students, the annual Stack Overflow Survey is a valuable resource for revealing the developer landscape. 2023 is the thirteenth year of this survey and it received 90,000 responses across 185 countries.

A new section on AI in the 2023 Developers Survey aimed to provide insights into both the usage of AI tools and what developers think about them. 

The headline finding is that 77% of respondents feel favourable towards using AI tools as part of their development workflow: 

Compared to the smaller survey we reported on last month, see What Devs Think About AI, Stack Overflow Developers are even more favourably disposed towards AI Tools.  However, only 42% of them trust the accuracy of the output of AI tools they use! 

 

The Stack Overflow survey distinguished between two categories of AI tools - Search Tools and Developer Tools.

 

ChatGPT emerged as top choice for AI Search with 83% of respondents having used it. Bing AI with 21%, WolframAlpha with 13% and Google Bard AI  with 10% were much less popular.

 

AI Developer Tools were less popular than AI Search tools. GitHub Copilot had been used by 55%, Tabnine by 13% and AWS CodeWhisperer by 5% 

 

Across all respondents to the survey 44% currently use AI tools, another 25% plan to do so soon and 29% don't use them and don't plan to. But looking in more detail reveals some interesting differences between those just learning to code and professional developers with different levels of experience.

Of those learning to code, who make up less than 6% of respondents, 55% currently use AI tools, 27% plan to do so soon and only 18% have no intention of using them. 

Professional developers account for 75% of respondents and 30% of them don't use or plan to use AI tools and the percentage increases with length of experience:

 

Stack Overflow Labs comment on this relationship is:

More experienced developers are typically working on more complex projects and/or code structures, things that AI tools aren’t able to help with right now. 

 

Stack Overflow Labs also finds that: 

Developers focused on hardware, backend systems, or applications are less likely to be using AI tools now. This same group doesn’t see themselves using these tools in the future either.

while: 

Data scientists, front-end, fullstack, and cloud infrastructure developers are more likely to be using the tools now. 

 

The survey asked the question:

Which parts of your development workflow are you currently using AI tools for and which are you interested in using AI tools for over the next year? Please select all that apply. 

 

The answers from developers who already used AI in their work gave very different answers from those planning to use them and those with no intention to do so:

 

 

 

The vast majority of developers actually using AI selected Writing code and half selected Debugging and getting help and just over a third selected Documenting code. 

For those planing to adopt AI tools soon Testing code came top (55%) with documenting code next (50%) while Writing code was the bottom use case (24%).

Writing Code was only selected as being useful by 4% of those not intending to use AI tools and Collaborating with teammates seen as the top use case, something which perhaps suggests a misunderstanding of Copilot which has been described as a virtual Pair Programmer.

Responses to the question:

Thinking about how your workflow and process changes over time, how similar or different do you anticipate your workflow to be 1 year from now as a result of AI tools you are currently using?

revealed that developers this that AI tools are going to have an impact in the way they work:

 

 

The areas where AI is most expected to make a difference are:

1. Learning about a codebase
2. Debugging and getting help 
3. Documenting code
4. Testing code

The fact that Writing Code isn't among the top use cases is presumably because for many developers that has already happened!

 

 

 

More Information

Stack Overflow 2023 Developer Survey

Related Articles 

What Devs Think About AI

Shifts & Stability In Developer Landscape

Stack Overflow Publishes Largest Ever Survey

 

To be informed about new articles on I Programmer, sign up for our weekly newsletter, subscribe to the RSS feed and follow us on Twitter, Facebook or Linkedin.

 

A Tee Is Not Just For Xmas - Top Tees 20/12/2024 Programmer gifts - easy idea, difficult implementation.  Here's our pick of tee-shirts for giving, buying or just wearing at any time of the year. + Full Story

Programmer Gifts - Pi For Xmas 13/12/2024 The holiday season is a good time to learn about computers - you have the time. But where to start? Our advice is to ignore the pudding and go for a Pi. + Full Story

More News AI At edX With 30% Savings GitHub Announces Open Source Security Fund Hour Of Code 2024 Is About To Kick Off Personal Picks For Holiday Gifts Rust 1.83 Improves Const Context Code Handling Zitadel Announces Funding And Future Plans Remembering Grace Hopper On Her 114th Anniversary Insights Into Learning Computer Science Eclipse IoT Developer Survey 2024 Pico 2W Announced But There Is A Surprise! Kafka 3.9 Adds Dynamic KRaft Quorums The Art Of Computer Programming - A Great Present Visual Studio 17.12 Released Along With Aspire

 

Comments

Make a Comment or View Existing Comments Using Disqus

or email your comment to: comments@i-programmer.info