--- title: "AI 2026" url: https://survey.stackoverflow.co/2026/ai year: 2026 chapter: "ai" licence: "ODbL 1.0 (https://opendatacommons.org/licenses/odbl/1-0/)" --- # AI 2026 How are technologists using AI? ## Highlights ### AI moves from on-the-side to built-in Coding assistants and agents have pulled ahead as the dominant use case for AI (66%), outpacing even general-purpose LLM-powered chatbots (63%). It’s a sign that AI has become less sidekick and more infrastructure, built right into the tools technologists already use. The overwhelming majority of respondents now use coding assistants and agents daily (73%), with nearly a third of daily users spending upwards of four hours. ### 31% of daily AI users use it for 4+ hours/day Of the over 70% of respondents who use AI daily, 31% use it over four hours per day. Asked as: You indicated that you use AI tools and/or agents daily. How much of your workday do you use AI tools to help you work? (`AIDaily`, single select, optional, v2026.1) All Respondents · n = 10,926 | | Respondents | Percent | | --- | ---: | ---: | | 4+ hours/day | 3,378 | 30.9% | | 1-2 hours/day | 2,765 | 25.3% | | 2-4 hours/day | 2,636 | 24.1% | | <1 hour every day | 1,335 | 12.2% | | <1 hour and not every day | 763 | 7.0% | | None | 49 | 0.5% | ### Despite plenty of societal concern about AI, survey respondents retain a fairly positive view of the technology. When asked about overall attitude toward AI tools, favorable views outnumber unfavorable ones by almost 3 to 1. But “favorable” doesn't mean “unconditionally positive.” Developers aren't simply handing AI the keys to important decisions. When asked when it’s acceptable to trust AI, 48% indicate they trust AI when they can easily validate the answers and only 16% extend that trust to most tasks that are not important work decisions. The sweet spot where most AI users land is trusting AI when they understand what it’s working on, such as with familiar code (67% currently use AI with this) or when debugging or refactoring (61% currently use AI with this). That’s why human-in-the-loop is the dominant mode of working with these tools. > “This is where I personally feel the importance of human judgment. If you leave it up to chance, if you leave it up to the roll of the dice of the LLM, you're not gonna get the specificity. Everything I've been doing and everything my team's been doing is imagining AI as an exoskeleton, like a Tony Stark augmentation, as opposed to an empty Ultron with nobody inside. Which nobody wants.” > > — [Scott Hanselman, VP of Developer Community at Microsoft](https://stackoverflow.blog/2026/01/13/vibe-code-anything-in-a-hanselminute/) ### The right tool for the job The same caution shows up in how people choose (and choose against) certain tools. Developers switching to new AI tools are doing it mostly because the output is better (25%) or because their organization required it (11%). ### 25% switch AI tools due to better output quality Seeking superior output quality is the single largest reason developers switch AI tools, though a third (34%) retained their original tool choices over the past year. For all the concerns over token costs, only 6% switched because of costs. Asked as: In the past year, have you switched from one AI tool to another for work? If yes, what was the primary reason? (`AIToolSwitch`, single select, optional, v2026.1) All Respondents · n = 13,144 | | Respondents | Percent | | --- | ---: | ---: | | No | 4,423 | 33.7% | | Yes — better output quality | 3,305 | 25.1% | | Yes — my organization required or recommended a different tool | 1,394 | 10.6% | | Yes — better integration with my workflow or development environment | 975 | 7.4% | | Yes — lower cost | 835 | 6.4% | | Yes — better support for agents, automation, or coding workflows | 487 | 3.7% | | Yes — another reason (write in) | 466 | 3.6% | | Yes — the previous tool became too expensive, limited, or unavailable | 433 | 3.3% | | Yes — better security, privacy, or compliance | 283 | 2.2% | | Yes — access to a preferred model or provider | 280 | 2.1% | | Yes — for a specific feature | 179 | 1.4% | | Not applicable — I do not use AI tools for work | 84 | 0.6% | What are they using? Claude Code (66%) and GitHub Copilot (59%) are the leaders for coding agents. Respondents are using LangChain (30%) and OpenAI Agents SDK (24%) for orchestration and Make (39%) and n8n (33%) for automation. AI memory and agent orchestration tools show a notable gap between current adoption and stated interest, suggesting that the category is still finding its footing. Coding agents, meanwhile, are a more mature market with clear leaders emerging. That selectivity becomes very clear once you look at where AI is actually being put to work. Basically, developers trust AI with machines but not with people. AI use is common, and commonly accepted, across code generation, debugging, testing, and understanding unfamiliar codebases. Even secondary review functions—analyzing telemetry, reviewing pull requests, hunting for security issues—are seeing meaningful AI adoption. ### The human difference But the moment the task shifts toward human communication or creative output—writing emails, generating multimedia, anything requiring a feel for what people want—usage falls off a cliff. The same split shows up in perceived usefulness: technical tasks like implementation and debugging are both useful or very useful to 70%+ of survey respondents, while communication and design lag behind, landing in “slop” territory with 35% and 49% respectively finding AI useful or very useful in those areas. ### AIWork_Current The top 5 tasks that respondents are currently using AI or AI agents for Asked as:

Which parts of your workflow are you currently integrating into AI or using AI agents to accomplish? Will they be more or less integrated over the next year? Please select one for each scenario.

(`AIWork_Current`) All Respondents · n = 12,547 | | Respondents | Percent | | --- | ---: | ---: | | Generating code in a familiar area | 8,402 | 69.3% | | Debugging, troubleshooting, or refactoring code | 7,684 | 63.8% | | Answering straightforward technical questions | 7,017 | 59.4% | | Writing or improving tests | 6,915 | 58.1% | | Generating code in an unfamiliar area | 6,707 | 56.0% | The inevitable takeaway is that technologists see AI as very good at technical problems with checkable answers, and considerably less good at solving the other everyday problems of work. A significant chunk of respondents who avoid AI completely do so out of concern for their own skills, privacy, or the technology’s enormous ethical and environmental footprint. ### AIAvoid 6 out of 10 users cite a reason to avoid using AI at work or school, losing job skills is a top concern. Asked as: Do you try to avoid using AI tools or agents at work or school? Select the option below that most accurately reflects your reasoning. (`AIAvoid`, single select, optional, v2026.1) All Respondents · n = 13,857 | | Respondents | Percent | | --- | ---: | ---: | | I do not avoid using AI at work or school | 5,728 | 41.3% | | Concern over losing job skills or training an AI to replace you | 2,367 | 17.1% | | Moral or ethical concerns | 2,108 | 15.2% | | Issues with privacy or security | 1,802 | 13.0% | | Environmental concerns | 1,129 | 8.2% | | Other, please specify: | 369 | 2.7% | | Difficulty or lack of training | 354 | 2.6% | Every figure in this chapter: https://survey.stackoverflow.co/2026/ai/data.md