AI Coding Assistant Statistics 2026
This page collects published figures on AI coding assistants such as GitHub Copilot, Claude Code, Cursor and Codex: how many developers use them and how often, which tools and agents they pick, how much shipped code is written by AI, what controlled studies and engineering telemetry say about productivity, and how much developers trust the output. Sources are developer surveys, engineering analytics firms, code review and security vendors, AI labs and academic field experiments. Most data covers 2025 and 2026, with a few 2024 studies kept because they remain the main randomized evidence.
By the DBotMaker research desk · Updated November 6, 2026 · 55 numbers drawn from 24 publishers and reports
Short answer
AI coding assistants are now used by almost every professional developer: 95% of 900 engineers in The Pragmatic Engineer's 2026 survey use AI tools at least weekly, and DX measured 52.7% of code as AI-authored in the second quarter of 2026, up from 24% in late 2025. Measured productivity gains are smaller than the adoption numbers suggest: a field experiment with 4,867 developers at Microsoft, Accenture and a third firm found 26% more completed tasks, and a Google trial found engineers about 21% faster, while Qodo found 76.4% of developers report frequent errors and low confidence in AI code.
Headline numbers
- 95%95% of about 900 software engineers surveyed by The Pragmatic Engineer in early 2026 used AI tools at least weekly, and 75% used AI for at least half of their engineering work.The Pragmatic Engineer
- 52.7%52.7% of code was AI-authored in DX's second-quarter 2026 data, up from 24% in the fourth quarter of 2025 and 34% in the first quarter of 2026.DX
- more than a quarterMore than a quarter of all new code at Google was generated by AI and then reviewed and accepted by engineers in late 2024, according to Alphabet's CEO.Google
- 26.08%Access to GitHub Copilot raised the weekly number of completed tasks by 26.08% across field experiments with 4,867 developers at Microsoft, Accenture and a third company.MIT Department of Economics (Cui, Demirer et al.)
- 21%Google engineers who used AI completed a realistic coding task about 21% faster than those who did not in a 2024 randomized controlled trial, though the result was not statistically significant.Google
- 98%Developers on teams with high AI adoption merged 98% more pull requests and completed 21% more tasks, but pull request review time rose 91%, in Faros AI telemetry from more than 10,000 developers in 2025.Faros AI
- 76.4%76.4% of 609 developers in Qodo's 2025 survey reported frequent AI hallucinations and low confidence in shipping AI-generated code.Qodo
- 1.7 timesAI-co-authored pull requests contained about 1.7 times more issues overall than human-only pull requests in CodeRabbit's analysis of 470 open-source GitHub pull requests in 2025.CodeRabbit
Reading the numbers
Adoption is close to universal among working developers, and the open question has moved from whether teams use AI to how much of the work it does. The Pragmatic Engineer found 95% of respondents using AI tools at least weekly in early 2026, DX put adoption across engineering organizations at 91% in late 2025, and Qodo found 82% using an assistant daily or weekly. Use has also shifted from autocomplete to agents: 55% of The Pragmatic Engineer's respondents regularly use AI agents, and Anthropic classified 79% of Claude Code conversations as automation rather than collaboration.
The share of code written by AI is rising quickly but is measured in very different ways. DX's quarterly reports show AI-authored code climbing from 24% in the fourth quarter of 2025 to 52.7% in the second quarter of 2026 across its customer sample. Google said in late 2024 that more than a quarter of its new code was generated by AI and then reviewed by engineers. Self-reports vary by seniority: Fastly found 32% of senior developers but only 13% of juniors say more than half of their shipped code is AI-generated.
Productivity evidence is positive but modest compared with vendor claims. The largest field experiment, by economists from MIT, Princeton and Microsoft across 4,867 developers, found a 26.08% increase in completed pull requests, and a randomized trial inside Google estimated engineers worked about 21% faster. Bain reported typical team gains of 10% to 15%, and DX's longitudinal analysis put most organizations at 5% to 15% more throughput. Telemetry from Faros AI shows where the gains go: individual developers merge 98% more pull requests, but review time rises 91%.
Trust and quality are the weak points. Qodo found 76.4% of developers fall into a high-hallucination, low-confidence group, Postman found 36% cite lack of trust in AI systems as an obstacle, and DX saw developers' change confidence fall 6.1% in one quarter of 2026. Code analysis from CodeRabbit found about 1.7 times more issues in AI-co-authored pull requests, and Apiiro measured 10 times more security findings from AI-assisted teams. Developers still like the tools: nearly 80% of Fastly's respondents say AI makes coding more enjoyable.
How many developers use AI coding assistants
- 1.
95% of about 900 software engineers surveyed by The Pragmatic Engineer in early 2026 used AI tools at least weekly, and 75% used AI for at least half of their engineering work.
The Pragmatic Engineer, AI Tooling for Software Engineers in 2026 (Mar 2026)
Exact wording at the source
95% of respondents use AI tools at least weekly, or more often, and 75% use AI for at least half their software engineering work.
- 2.
Adoption of AI coding assistants reached 91% across engineering organizations in DX's fourth-quarter 2025 data.
DX, AI-assisted engineering: Q4 impact report (Jan 2026)
Exact wording at the source
The use of AI coding assistants is now ubiquitous in engineering organizations, with overall adoption reaching 91%.
- 3.
82% of 609 developers surveyed by Qodo in 2025 used an AI coding assistant daily or weekly.
Qodo, State of AI Code Quality in 2025 (Jun 2025)
Exact wording at the source
82% say they use an AI coding assistant daily or weekly, a clear sign that these tools have moved from experimentation to core workflow.
- 4.
89% of more than 5,700 developers, architects and executives in Postman's 2025 survey used AI, but only 24% designed APIs with AI agents in mind.
Postman, 2025 State of the API Report (Aug 2025)
Exact wording at the source
89% of developers use AI, but only 24% design APIs for AI agents.
- 5.
More than 97% of 2,000 enterprise software staff in the US, Brazil, India and Germany had used AI coding tools at work at some point by early 2024, according to GitHub.
GitHub, Survey: The AI wave continues to grow on software development teams (Aug 2024)
Exact wording at the source
More than 97% of respondents reported having used AI coding tools at work at some point, a finding consistent across all four countries.
- 6.
76% of IT and SaaS professionals in Docker's 2025 State of Application Development survey used AI tools at work, compared with 22% across all industries.
Docker, 2025 Docker State of Application Development Report (Jul 2025)
Exact wording at the source
IT is among the leaders in AI — with 76% of IT/SaaS pros using AI tools at work, compared with just 22% across industries.
- 7.
80% of developers at companies with fewer than 50 developers, and just over 70% at companies with 50 to 200 developers, used AI at least weekly in DX's late 2025 data.
DX, AI-assisted engineering: Q4 impact report (Jan 2026)
Exact wording at the source
Eighty percent of developers at companies with fewer than 50 developers, and just over 70% of developers at companies with 50-200 developers, are using AI at least weekly.
- 8.
In field experiments at Microsoft, Accenture and a third company in 2023 and 2024, around 30% to 40% of engineers given GitHub Copilot never tried it.
MIT Department of Economics (Cui, Demirer et al.), The Effects of Generative AI on High-Skilled Work: Evidence from Three Field Experiments with Software Developers (2025)
Exact wording at the source
Despite this, the adoption rate is significantly below 100% in all three experiments, with around 30-40% of the engineers not even trying the product.
- 9.
Two out of three software companies had rolled out generative AI tools by 2025, but developer adoption inside those firms was low, according to Bain.
Bain & Company, From Pilots to Payoff: Generative AI in Software Development (Technology Report 2025) (Sep 2025)
Exact wording at the source
Two out of three software firms have rolled out generative AI tools, and among those, developer adoption is low.
Which assistants and agents developers use
- 10.
55% of engineers in The Pragmatic Engineer's 2026 survey regularly used AI agents, rising to 63.5% among staff-level and more senior engineers.
The Pragmatic Engineer, AI Tooling for Software Engineers in 2026 (Mar 2026)
Exact wording at the source
AI agent usage rising : 55% of respondents now regularly use AI agents, with staff+ engineers leading adoption on 63.5% usage in the survey results.
- 11.
Among engineers who regularly use AI agents, 71% used Claude Code, 46% GitHub Copilot and 39% Cursor in early 2026.
The Pragmatic Engineer, AI Tooling for Software Engineers in 2026 (Mar 2026)
Exact wording at the source
Claude Code is the clear leader (71% usage), followed by GitHub Copilot (46%) and Cursor (39%)
- 12.
56% of engineers at companies with 10,000 or more staff used GitHub Copilot in 2026, while 75% at the smallest companies used Claude Code.
The Pragmatic Engineer, AI Tooling for Software Engineers in 2026 (Mar 2026)
Exact wording at the source
Huge companies (10K+) more likely to use Copilot (56%), tiny startups mostly go with Claude Code (75%) and Cursor (42%).
- 13.
59% of developers in Qodo's 2025 survey used three or more AI tools regularly, and 20% used five or more.
Qodo, State of AI Code Quality in 2025 (Jun 2025)
Exact wording at the source
59% use three or more AI tools regularly, and 20% manage five or more
- 14.
49% of 3,266 DevSecOps professionals surveyed for GitLab in 2025 used more than five AI tools, and 60% used more than five tools for software development overall.
GitLab, GitLab Survey Reveals the AI Paradox (Nov 2025)
Exact wording at the source
60% use more than five tools for software development; 49% use more than five AI tools
- 15.
70% of developers in Postman's 2025 survey were aware of the Model Context Protocol, but only 10% used it regularly.
Postman, 2025 State of the API Report (Aug 2025)
Exact wording at the source
While 70% of developers are aware of MCP, only 10% are using it regularly, pointing to a growing interest but limited readiness.
- 16.
79% of Claude Code conversations analyzed by Anthropic in early 2025 were automation, where the AI performs the task directly, compared with 49% on Claude.ai.
Anthropic, Anthropic Economic Index: AI's impact on software development (Apr 2025)
Exact wording at the source
Here, we found that Claude Code showed dramatically higher automation rates—79% of conversations involved some form of automation, compared to 49% on Claude.ai.
How much code AI writes
- 17.
52.7% of code was AI-authored in DX's second-quarter 2026 data, up from 24% in the fourth quarter of 2025 and 34% in the first quarter of 2026.
DX, The quality paradox of AI-generated code (Aug 2026)
Exact wording at the source
The Q2 report found that 52.7% of code is AI-authored, up from 24% in Q4 2025 and 34% in Q1 2026.
- 18.
Daily users of AI coding assistants reported that 24% of their merged code was AI-generated in late 2025, compared with just over 20% for monthly users.
DX, AI-assisted engineering: Q4 impact report (Jan 2026)
Exact wording at the source
Daily users of AI coding assistants report that 24% of their merged code is AI-generated, with monthly users reporting just over 20%.
- 19.
More than a quarter of all new code at Google was generated by AI and then reviewed and accepted by engineers in late 2024, according to Alphabet's CEO.
Google, Q3 earnings call: CEO's remarks (Oct 2024)
Exact wording at the source
Today, more than a quarter of all new code at Google is generated by AI, then reviewed and accepted by engineers.
- 20.
32% of senior developers with 10 or more years of experience said over half of their shipped code was AI-generated in Fastly's July 2025 survey of 791 developers.
Fastly, Why Senior Developers Ship 2.5x More AI-Generated Code (Aug 2025)
Exact wording at the source
By contrast, 32% of senior developers say the same, suggesting that more experienced engineers are not only using AI more aggressively, but are also trusting it more in production environments.
- 21.
13% of junior developers with up to two years of experience said over half of their shipped code was AI-generated in 2025.
Fastly, Why Senior Developers Ship 2.5x More AI-Generated Code (Aug 2025)
Exact wording at the source
Among junior devs, just 13% say over half of their shipped code is AI-generated.
- 22.
65% of developers in Qodo's 2025 survey said at least a quarter of each commit was generated or shaped by AI.
Qodo, State of AI Code Quality in 2025 (Jun 2025)
Exact wording at the source
65 % of developers say at least a quarter of each commit is generated or shaped by AI.
- 23.
15% of developers surveyed by Qodo in 2025 said more than 80% of their code was AI-influenced.
Qodo, State of AI Code Quality in 2025 (Jun 2025)
Exact wording at the source
15 % say that more than 80% of their code is now AI-influenced.
- 24.
56% of engineers in The Pragmatic Engineer's 2026 survey did 70% or more of their engineering work using AI.
The Pragmatic Engineer, AI Tooling for Software Engineers in 2026 (Mar 2026)
Exact wording at the source
56% do 70% or more of their work using AI
- 25.
Mostly coding-related tasks made up 34% of Claude.ai conversations in November 2025, down from a peak of 40% in March 2025, according to Anthropic.
Anthropic, Anthropic Economic Index report: January 2026 (Jan 2026)
Exact wording at the source
Such dominance has subsided on Claude.ai: the share of conversations on Claude.ai assigned to such (mostly) coding-related tasks is down from a peak of 40% in March 2025 to 34% in November 2025.
Chart
Developers who say over half their shipped code is AI-generated
Survey of 791 professional developers by Fastly, July 2025.
- Junior developers (0 to 2 years)13%
- Senior developers (10+ years)32%
Source: Fastly, Why Senior Developers Ship 2.5x More AI-Generated Code
What productivity studies find
- 26.
Access to GitHub Copilot raised the weekly number of completed tasks by 26.08% across field experiments with 4,867 developers at Microsoft, Accenture and a third company.
MIT Department of Economics (Cui, Demirer et al.), The Effects of Generative AI on High-Skilled Work: Evidence from Three Field Experiments with Software Developers (2025)
Exact wording at the source
Our preferred estimates from an instrumental variable regression suggest that usage of the coding assistant causes a 26.08% (SE: 10.3%) increase in the weekly number of completed tasks for those using the tool.
- 27.
The same experiments found weekly builds rose 38.38% and weekly commits rose 13.55% among developers using Copilot.
MIT Department of Economics (Cui, Demirer et al.), The Effects of Generative AI on High-Skilled Work: Evidence from Three Field Experiments with Software Developers (2025)
Exact wording at the source
on average, the number of weekly pull requests made by developers increases by 26.08% (SE: 10.3%), the number of weekly commits increases by 13.55% (SE: 10.0%), and the number of weekly builds increases by 38.38% (SE: 12.55%).
- 28.
Google engineers who used AI completed a realistic coding task about 21% faster than those who did not in a 2024 randomized controlled trial, though the result was not statistically significant.
Google, How much does AI impact development speed? An enterprise-based randomized controlled trial (Oct 2024)
Exact wording at the source
Our study shows that developers who used AI were about 21% faster than those who did not, controlling for other factors.
- 29.
Software teams using AI coding assistants saw 10% to 15% productivity gains in 2025, according to Bain, which found the saved time was often not redirected to higher-value work.
Bain & Company, From Pilots to Payoff: Generative AI in Software Development (Technology Report 2025) (Sep 2025)
Exact wording at the source
Teams using AI assistants see 10% to 15% productivity boosts, but often the time saved isn’t redirected toward higher-value work.
- 30.
US workers in computer and mathematical occupations used generative AI in nearly 12% of their work hours in late 2024 and said it saved 2.5% of their work time.
Federal Reserve Bank of St. Louis, The Impact of Generative AI on Work Productivity (Feb 2025)
Exact wording at the source
Workers in the computer and mathematics occupation used generative AI in nearly 12% of their work hours, and they reported this saved them 2.5% of work time.
- 31.
Across more than 400 companies tracked from November 2024 to February 2026, DX found most organizations gained 5% to 15% in pull request throughput from AI coding tools.
DX, AI and engineering velocity: a longitudinal analysis (Mar 2026)
Exact wording at the source
For most organizations, a 5–15% throughput gain is what the current generation of AI coding tools is delivering.
- 32.
Developers in DX's second-quarter 2026 sample saved over 6 hours a week on average with AI, but the saved time was not yet visibly turning into new features.
DX, State of AI Impact in Engineering: Q2 Report (Jul 2026)
Exact wording at the source
Time savings are real and growing, now over 6 hours a week on average, but saved hours are not yet visibly converting into new value creation at the portfolio level.
- 33.
Weekly pull request throughput rose 75.5% over the four quarters to September 2026 across more than 500 companies in DX's data.
DX, State of AI Impact in Engineering: Q3 Report (Oct 2026)
Exact wording at the source
Organizations generating high volumes of AI code wait 105% longer for a first review than those at the bottom, yet weekly PR throughput keeps climbing fast, up 75.5% over four quarters.
- 34.
Developers on teams with high AI adoption merged 98% more pull requests and completed 21% more tasks, but pull request review time rose 91%, in Faros AI telemetry from more than 10,000 developers in 2025.
Faros AI, The AI Productivity Paradox Report 2025 (Jul 2025)
Exact wording at the source
Developers on teams with high AI adoption complete 21% more tasks and merge 98% more pull requests , but PR review time increases 91% , revealing a critical bottleneck: human approval.
- 35.
59% of senior developers said AI tools helped them ship faster overall in 2025, compared with 49% of junior developers.
Fastly, Why Senior Developers Ship 2.5x More AI-Generated Code (Aug 2025)
Exact wording at the source
Even so, 59% of senior developers say AI tools help them ship faster overall, compared to 49% of juniors.
Chart
Change on teams with high AI adoption
Telemetry from 1,255 teams and more than 10,000 developers across multiple companies, 2025.
- Tasks completed21%
- Pull requests merged98%
- Pull request review time91%
Source: Faros AI, The AI Productivity Paradox Report 2025
How much developers trust AI-generated code
- 36.
76.4% of 609 developers in Qodo's 2025 survey reported frequent AI hallucinations and low confidence in shipping AI-generated code.
Qodo, State of AI Code Quality in 2025 (Jun 2025)
Exact wording at the source
Bottom Right (76.4%) High hallucinations, low confidence This is the dominant group.
- 37.
25% of developers surveyed by Qodo in 2025 estimated that 1 in 5 AI-generated suggestions contain factual errors or misleading code.
Qodo, State of AI Code Quality in 2025 (Jun 2025)
Exact wording at the source
25% of developers estimate that 1 in 5 AI-generated suggestions contain factual errors or misleading code.
- 38.
65% of developers using AI for refactoring said the assistant misses relevant context, the most common complaint in Qodo's 2025 survey.
Qodo, State of AI Code Quality in 2025 (Jun 2025)
Exact wording at the source
65% of developers using AI for refactoring and ~60% for testing, writing, or reviewing say the assistant “misses relevant context.”
- 39.
36% of developers in Postman's 2025 survey named lack of trust in AI systems as a main obstacle to using AI tools at work, and 33% cited ethical, legal or compliance concerns.
Postman, 2025 State of the API Report (Aug 2025)
Exact wording at the source
When asked about the biggest obstacles to using AI tools at their organizations, developers cite real trust and compliance issues: 36% lack trust in AI systems, and 33% have ethical, legal, and compliance concerns.
- 40.
28% of developers in Fastly's 2025 survey said they frequently have to fix or edit AI-generated code enough to cancel out most of the time saved.
Fastly, Why Senior Developers Ship 2.5x More AI-Generated Code (Aug 2025)
Exact wording at the source
Nearly 1 in 3 developers (28%) say they frequently have to fix or edit AI-generated code enough that it offsets most of the time savings.
- 41.
Developers' change confidence, their trust that a change will not break things, fell 6.1% between the first and second quarters of 2026 across more than 500 companies measured by DX.
DX, The quality paradox of AI-generated code (Aug 2026)
Exact wording at the source
Across a sample of more than 500 companies, code maintainability improved 3.8% between Q1 and Q2. Over the same period change confidence fell 6.1%.
- 42.
90% of US enterprise developers in GitHub's 2024 survey reported a perceived increase in code quality from AI coding tools, compared with 60% in Germany.
GitHub, Survey: The AI wave continues to grow on software development teams (Aug 2024)
Exact wording at the source
Most respondents in the U.S. (90%) and India (81%), along with more than half in Brazil (61%) and Germany (60%), reported a perceived increase in code quality when using AI coding tools.
Chart
Developers reporting a perceived increase in code quality from AI tools
Survey of 2,000 enterprise software staff, 500 per country, conducted by Wakefield Research for GitHub, February to March 2024.
- United States90%
- India81%
- Brazil61%
- Germany60%
Source: GitHub, Survey: The AI wave continues to grow on software development teams
Code quality and security of AI-written code
- 43.
AI-co-authored pull requests contained about 1.7 times more issues overall than human-only pull requests in CodeRabbit's analysis of 470 open-source GitHub pull requests in 2025.
CodeRabbit, State of AI vs Human Code Generation Report (Dec 2025)
Exact wording at the source
AI-generated PRs contained ~1.7× more issues overall.
- 44.
Logic and correctness issues were 75% more common in AI-co-authored pull requests than in human-only ones, according to CodeRabbit.
CodeRabbit, State of AI vs Human Code Generation Report (Dec 2025)
Exact wording at the source
Logic and correctness issues were 75% more common in AI PRs.
- 45.
AI-assisted teams in Fortune 50 enterprises shipped 10 times more security findings than their peers in Apiiro's 2025 code analysis.
Apiiro, 4x Velocity, 10x Vulnerabilities: AI Coding Assistants Are Shipping More Risks (Sep 2025)
Exact wording at the source
AI-assisted teams didn’t just ship faster – they shipped 10× more security findings.
- 46.
By June 2025, AI-generated code was adding over 10,000 new security findings per month across the Fortune 50 repositories Apiiro studied, 10 times the December 2024 level.
Apiiro, 4x Velocity, 10x Vulnerabilities: AI Coding Assistants Are Shipping More Risks (Sep 2025)
Exact wording at the source
By June 2025, AI-generated code was introducing over 10,000 new security findings per month across the repositories in our study — a 10× spike in just six months compared to December 2024.
- 47.
Privilege escalation paths in code rose 322% and architectural design flaws 153% as AI-written code spread, while trivial syntax errors fell, according to Apiiro.
Apiiro, 4x Velocity, 10x Vulnerabilities: AI Coding Assistants Are Shipping More Risks (Sep 2025)
Exact wording at the source
Privilege escalation paths jumped 322%, and architectural design flaws spiked 153%.
- 48.
Higher AI adoption was associated with a 9% increase in bugs per developer and a 154% increase in average pull request size in Faros AI's 2025 telemetry.
Faros AI, The AI Productivity Paradox Report 2025 (Jul 2025)
Exact wording at the source
AI adoption is consistently associated with a 9% increase in bugs per developer and a 154% increase in average PR size.
- 49.
Developers with GitHub Copilot access introduced 41% more bugs than a comparison group without it in Uplevel's 2024 study of engineering teams.
Uplevel, Can GenAI Actually Improve Developer Productivity? (Oct 2024)
Exact wording at the source
The group using Copilot introduced 41% more bugs, suggesting copilot might affect code quality, given it didn’t affect PR throughput.
- 50.
Claude judged software development requests on Claude.ai as successfully completed 61% of the time in Anthropic's November 2025 sample, compared with 78% for personal tasks.
Anthropic, Anthropic Economic Index report: January 2026 (Jan 2026)
Exact wording at the source
Claude assesses personal tasks as successfully completed 78% of the time, versus 61% for software development.
Developer sentiment, jobs and spending
- 51.
Nearly 80% of developers in Fastly's 2025 survey said AI tools make coding more enjoyable.
Fastly, Why Senior Developers Ship 2.5x More AI-Generated Code (Aug 2025)
Exact wording at the source
Productivity gains may be murky, but job satisfaction isn't: nearly 80% of developers say AI tools make coding more enjoyable.
- 52.
76% of DevSecOps professionals surveyed for GitLab in 2025 believed that as AI makes coding easier there will be more engineers, not fewer.
GitLab, GitLab Survey Reveals the AI Paradox (Nov 2025)
Exact wording at the source
With 76% of respondents believing that as coding gets easier with AI, there will be more engineers rather than fewer
- 53.
99% to 100% of enterprise developers in GitHub's 2024 four-country survey believed AI skills make them more attractive job candidates.
GitHub, Survey: The AI wave continues to grow on software development teams (Aug 2024)
Exact wording at the source
Nearly all respondents (99-100%) believe this skill makes them more attractive candidates, underlining the growing importance of AI across various fields.
- 54.
21% of US technology industry employees thought it likely that AI, automation or robots would eliminate their job within five years, among the highest of any industry in Gallup's 2025 survey.
Gallup, AI Use at Work Has Nearly Doubled in Two Years (Jun 2025)
Exact wording at the source
Some industries — such as technology (21%), retail (21%) and finance (20%) — are more likely than others to believe AI will eliminate their job.
- 55.
Median quarterly AI spend at technology companies grew from about $1,500 to about $44,000 in a year, roughly 28 times, in DX's second-quarter 2026 data.
DX, State of AI Impact in Engineering: Q2 Report (Jul 2026)
Exact wording at the source
Median quarterly AI spend grew from ~$1.5K to ~$44K in a year, roughly a 28x increase in the Tech sector alone
Questions readers ask
- What percentage of developers use AI coding assistants?
- Almost all professional developers now use them. The Pragmatic Engineer found 95% of about 900 engineers using AI tools at least weekly in early 2026, DX measured 91% adoption across engineering organizations in late 2025, and Qodo found 82% of developers using an assistant daily or weekly.
- How much code is written by AI?
- DX measured 52.7% of code as AI-authored in the second quarter of 2026, up from 24% in the fourth quarter of 2025. Google said in late 2024 that more than a quarter of its new code was generated by AI and reviewed by engineers, and Fastly found 32% of senior developers say over half their shipped code is AI-generated.
- Do AI coding assistants make developers more productive?
- Controlled studies show moderate gains. Field experiments with 4,867 developers found a 26.08% increase in completed tasks with GitHub Copilot, and a Google trial estimated engineers were about 21% faster. Bain reported typical team gains of 10% to 15%, and DX found most organizations gained 5% to 15% in throughput.
- Do developers trust AI-generated code?
- Many do not. Qodo found 76.4% of developers report frequent hallucinations and low confidence in shipping AI code, Postman found 36% name lack of trust in AI systems as an obstacle, and Fastly found 28% often spend enough time fixing AI output to cancel most of the time saved.
- Is AI-generated code less secure?
- Code analysis studies point that way. Apiiro found AI-assisted teams in Fortune 50 companies shipped 10 times more security findings, and CodeRabbit found AI-co-authored pull requests had about 1.7 times more issues overall and 75% more logic and correctness issues than human-only ones.
- Which AI coding assistant is most used?
- It depends on company size. In The Pragmatic Engineer's 2026 survey, 71% of regular agent users used Claude Code, 46% GitHub Copilot and 39% Cursor. At companies with 10,000 or more staff, GitHub Copilot led at 56%, while 75% of engineers at the smallest companies used Claude Code.
Where these numbers come from
We selected figures from the organizations that collected the data: developer surveys run by tool makers and publishers, engineering analytics firms reporting telemetry from their customers, code review and security companies analyzing repositories, AI labs analyzing their own usage, a Federal Reserve bank and academic field experiments. Each quote was checked against the source page on the date shown. Samples and definitions differ a lot: some surveys reach engaged readers or customers rather than all developers, vendors publish research that relates to their own products, and terms such as AI-authored code, adoption and productivity are measured in different ways. Read figures from different sources as separate data points rather than a single trend. A few 2024 studies are included because they remain the main randomized evidence on coding assistants. We store the sentence each number appears in, and a script loads the publisher's page again to make sure that sentence is still there. Our methodology has the details.
Sources
- Anthropic. Anthropic Economic Index: AI's impact on software development (Apr 2025).
- Anthropic. Anthropic Economic Index report: January 2026 (Jan 2026).
- Apiiro. 4x Velocity, 10x Vulnerabilities: AI Coding Assistants Are Shipping More Risks (Sep 2025).
- Bain & Company. From Pilots to Payoff: Generative AI in Software Development (Technology Report 2025) (Sep 2025).
- CodeRabbit. State of AI vs Human Code Generation Report (Dec 2025).
- Docker. 2025 Docker State of Application Development Report (Jul 2025).
- DX. AI-assisted engineering: Q4 impact report (Jan 2026).
- DX. The quality paradox of AI-generated code (Aug 2026).
- DX. AI and engineering velocity: a longitudinal analysis (Mar 2026).
- DX. State of AI Impact in Engineering: Q2 Report (Jul 2026).
- DX. State of AI Impact in Engineering: Q3 Report (Oct 2026).
- Faros AI. The AI Productivity Paradox Report 2025 (Jul 2025).
- Fastly. Why Senior Developers Ship 2.5x More AI-Generated Code (Aug 2025).
- Federal Reserve Bank of St. Louis. The Impact of Generative AI on Work Productivity (Feb 2025).
- Gallup. AI Use at Work Has Nearly Doubled in Two Years (Jun 2025).
- GitHub. Survey: The AI wave continues to grow on software development teams (Aug 2024).
- GitLab. GitLab Survey Reveals the AI Paradox (Nov 2025).
- Google. Q3 earnings call: CEO's remarks (Oct 2024).
- Google. How much does AI impact development speed? An enterprise-based randomized controlled trial (Oct 2024).
- MIT Department of Economics (Cui, Demirer et al.). The Effects of Generative AI on High-Skilled Work: Evidence from Three Field Experiments with Software Developers (2025).
- Postman. 2025 State of the API Report (Aug 2025).
- Qodo. State of AI Code Quality in 2025 (Jun 2025).
- The Pragmatic Engineer. AI Tooling for Software Engineers in 2026 (Mar 2026).
- Uplevel. Can GenAI Actually Improve Developer Productivity? (Oct 2024).
Citing this page
Link to this page or to a single numbered figure. If you reuse a number, name the organization that released it in the first place.
DBotMaker. (2026, November 6). AI Coding Assistant Statistics 2026. https://dbotmaker.io/ai-coding-assistant-statistics
Found a number that changed or a source that moved? Email editorial@dbotmaker.io. We plan to check this page again by January 5, 2027.