Developer productivity is being reshaped by AI tools, but the 2024 evidence also points to persistent friction in finding answers, handling technical debt, testing, and securing software. The figures below describe named survey populations and research groups; they are not universal measurements of every developer.
Contents
- AI adoption and perceived productivity
- Time allocation and workflow friction
- Developer experience measurement
- Delivery speed and engineering throughput
- Code quality, testing, and security
- Collaboration, learning, and sustainable productivity
AI adoption and perceived productivity
The AI section of the 2024 Stack Overflow Developer Survey shows broad exposure to AI-assisted development. In the 2024 survey, 76% of respondents were using or planning to use AI tools in development, compared with 70% in 2023. The share already using AI rose from 44% in 2023 to 62% in 2024. Overall, 72% viewed development AI tools favorably or very favorably.
Productivity was the leading motivation. Eighty-one percent of respondents identified increasing productivity as the biggest benefit they sought from AI tools. The learning effect was also prominent: 71% of respondents learning to code considered speeding up learning an important benefit, compared with 61% of professional developers.
Adoption does not mean uncritical trust. Forty-three percent felt good about the accuracy of AI output, while 31% were skeptical. Among professional developers, 45% rated AI tools bad or very bad at handling complex tasks. The most common current use was writing code, reported by 82% of respondents who were already using AI tools. Among respondents interested in AI but not yet using it, 46% were most interested in testing code.
The employment picture was comparatively measured: 70% of professional developers did not perceive AI as a threat to their current job. In the same survey, 79% identified misinformation or disinformation in AI results as a leading ethical concern, and 65% identified source attribution as a leading concern.
Time allocation and workflow friction
Finding information is a measurable part of the workday. According to the Professional Developers section of the 2024 Stack Overflow Developer Survey, 70% of professional developers knew where to find answers at work, but only 56% said they could find those answers quickly. Waiting for answers disrupted the workflow for 53%.
Knowledge silos were especially visible in the frequency data. Thirty percent of professional developers said knowledge silos affected their productivity at least 10 times per week, while 45% agreed or strongly agreed that knowledge silos were a nuisance. Sixty-one percent spent more than 30 minutes per day searching for answers or solutions. The same proportion of professional developer people managers spent more than 30 minutes per day answering workplace questions.
The first destination for technical questions was still a traditional public search engine for 55% of professional developers. AI-powered search was the first choice for 15%. Technical debt was the top workplace frustration for 63% of professional developers. These figures suggest that productivity measurement needs to include waiting, searching, and maintenance work—not only time spent producing new code.
The survey also provides context about the population behind these numbers: 41% of professional developers who answered the industry question worked in the software development industry. The findings are self-reported survey results, not time-tracking measurements across all workplaces.
Developer experience measurement
Developer productivity is increasingly treated as an experience and systems question. In the 2024 JetBrains State of Developer Ecosystem Report, approximately 50% of technology managers said their companies measured developer productivity, developer experience, or both. Only 16% of companies had dedicated developer productivity engineering or developer experience specialists.
The Stack Overflow professional-developer population was largely made up of individual contributors: 87% were individual contributors rather than people managers. Twenty-five percent reported one to four years of professional work experience. These details matter when comparing productivity signals, because a measure that is useful for an experienced individual contributor may not describe a manager’s work or a newer developer’s learning curve.
Work environments were predominantly cloud-based or hybrid. Eighty-six percent of professional developers worked in cloud-hosted or hybrid environments, while 14% worked solely with on-premises hosting. Access to AI-assisted technology at work reached 32.4% in 2024, compared with 15.7% in the prior Stack Overflow survey year.
The results support a practical measurement principle: track the conditions that make delivery possible. Access to tools, searchable knowledge, manageable technical debt, and the time spent unblocking colleagues are part of developer experience even when they do not appear as lines of code or completed tickets.
Delivery speed and engineering throughput
Executive-level results show strong expectations for productivity gains, alongside uneven implementation. In GitLab’s April 2024 survey, 99% of surveyed CxOs said developer productivity could help their business in some way. Sixty-nine percent said their organizations were shipping software at least twice as fast as a year earlier. Yet only 26% reported that their organization was implementing AI for software development, and 17% had already begun consolidating their software-development toolchains.
The distribution of engineering time helps explain why speed cannot be reduced to code generation. GitLab reported that only 25% of developers’ time was spent on code generation in its 2023 survey. GitHub’s enterprise research with Accenture found that developers could code up to 55% faster with the AI pair programmer. That result is an enterprise research finding and should not be generalized to every AI tool, team, or codebase.
GitHub’s 2024 enterprise survey also captured expectations about customer responsiveness. Seventy-three percent of U.S. respondents expected AI coding tools to improve their ability to meet customer requirements, compared with 61% of German respondents. In both the U.S. and Germany, 47% said time saved by AI coding tools was used for collaboration and system design. This is a useful distinction: faster implementation can create capacity for higher-level engineering work rather than simply increasing the volume of code.
| Productivity signal | Reported figure | Population and period |
|---|---|---|
| Organizations shipping at least twice as fast | 69% | GitLab-surveyed global CxOs, April 2024 |
| Developers’ time spent on code generation | 25% | GitLab survey, 2023 |
| Potential coding speed increase | Up to 55% | GitHub enterprise Copilot research, 2024 |
| Time saved used for collaboration and system design | 47% | U.S. and German GitHub respondents, 2024 |
Code quality, testing, and security
Speed is only useful when quality work keeps pace. GitHub’s 2024 enterprise survey found that more than 98% of respondents said their organizations had experimented with AI-generated test cases. Ninety-two percent of U.S. respondents said their organizations used AI for test generation at least sometimes, compared with 65% of German respondents.
Perceived code-quality improvement varied by country. It was reported by 90% of U.S. respondents, 81% of Indian respondents, 61% of Brazilian respondents, and 60% of German respondents. These are perceptions from the survey populations, not independent quality audits or defect-rate measurements.
The earlier GitLab Global DevSecOps Report shows how limited the code-generation slice can be. In its 2023 survey, developers spent 11% of their time testing code and 7% identifying and mitigating security vulnerabilities. Together, those activities represented a measurable workload beyond generation. A productivity program that counts only generated code can therefore miss testing and security work that protects delivery quality.
For C programming teams, the same measurement logic applies to build reliability, test coverage, memory-safety review, dependency management, and security analysis, but the supplied survey figures do not quantify those C-specific activities. The reported numbers support the broader point about balancing throughput with verification; they do not establish a separate benchmark for C developers.
Collaboration, learning, and sustainable productivity
GitHub’s 2023 survey on AI’s impact on the developer experience found that 81% of U.S. enterprise respondents expected AI coding tools to increase team collaboration. Fifty-seven percent said the tools helped improve their coding-language skills, and 41% believed they could help prevent burnout. In earlier GitHub Copilot research reported in 2023, 87% of respondents said the tool helped preserve mental effort on repetitive tasks.
These benefits coexist with learning pressure. Thirty percent of the 2023 U.S. enterprise respondents said learning and development negatively affected their overall workday. Approximately 33% said managers measured performance by code volume. The contrast is important: code volume can be easy to count, while preserved mental effort, skill development, and collaboration are harder to observe but directly connected to sustainable work.
Stack Overflow’s 2024 survey also recorded expectations about where AI would become more integrated within one year. Eighty-one percent expected more integration into code documentation, 80% into code testing, and 76% into code writing. These are forecasts from respondents, not measured future outcomes.
The evidence points toward a balanced dashboard: adoption and perceived usefulness, time spent searching and waiting, technical-debt friction, delivery speed, testing and security effort, collaboration, and learning. It should also preserve scope labels. The GitHub 2024 enterprise survey covered 2,000 respondents—500 each in the U.S., Brazil, Germany, and India. Its 2023 survey covered 500 U.S.-based developers at companies with 1,000 or more employees. Stack Overflow and GitLab populations were broader but self-reported. Those boundaries should remain attached to the numbers whenever they are used to guide decisions.