Claim analyzed

Tech

“Artificial intelligence is responsible for generating the majority of software code being written as of 2026.”

The conclusion

Reviewed by Vicky Dodeva, editor · Apr 03, 2026
False
2/10

The claim that AI generates the majority of software code as of 2026 is not supported by the evidence. The most rigorous measurements place AI-authored code at 22–29% of actual code output, while the often-cited 41% figure from JetBrains refers to lines "touched" by AI — not independently generated. High adoption rates for AI coding tools do not equate to AI writing most code. No credible primary dataset shows AI-generated code exceeding 50% globally.

Based on 20 sources: 4 supporting, 4 refuting, 12 neutral.

Caveats

  • The claim conflates 'AI-assisted' or 'AI-touched' code with 'AI-generated' code — a critical distinction. JetBrains' 41% figure explicitly refers to lines touched by AI, not autonomously written.
  • Supporting arguments rely heavily on a self-selected newsletter audience (The Pragmatic Engineer) skewed toward AI enthusiasts, which is not representative of the global developer population.
  • High AI tool adoption rates (85%+) do not logically entail that AI produces the majority of code — usage frequency and code volume share are separate metrics.

Sources

Sources used in the analysis

#1
Stack Overflow 2025-06-01 | AI | 2025 Stack Overflow Developer Survey
NEUTRAL

ChatGPT (82%) and GitHub Copilot (68%) are the clear market leaders, serving as the primary entry point for most developers using out-of-the-box AI assistance. Conversely to usage, positive sentiment for AI tools has decreased in 2025: 70%+ in 2023 and 2024 to just 60% this year.

#2
JetBrains 2025-11 | JetBrains Developer Ecosystem 2025
NEUTRAL

Survey of 24,000+ developers: 85% use AI tools regularly, but only about 30-40% of code is AI-assisted on average; 41% figure refers to lines touched by AI, not fully generated independently.

#3
Trending Topics 2025-12-31 | AI-Written Code Surges to 29% in US Software Development, Study Reveals
REFUTE

The share of AI-written code in the USA has risen from 5 percent in 2022 to 29 percent by the end of 2025. The research team analyzed more than 30 million Python code contributions from around 160,000 developers on GitHub.

#4
JetBrains 2025-10-01 | The State of Developer Ecosystem 2025: Coding in the Age of AI
NEUTRAL

AI is becoming a standard in developers' lives: 85% of developers regularly use AI tools for coding and development, and 62% rely on at least one AI tool as an essential part of their workflow.

#5
Grand View Research 2025-01-01 | AI Code Tools Market Size & Share | Industry Report, 2030
NEUTRAL

The global ai code tools market size was estimated at USD 4.86 billion in 2023 and is projected to reach USD 26.03 billion by 2030, growing at a CAGR of 27.1% from 2024 to 2030. The tools segment led the market in 2023, accounting for over 76.0% share of the global revenue.

#6
The Pragmatic Engineer 2026-01-01 | AI Tooling for Software Engineers in 2026
SUPPORT

95% of respondents report using AI tools at least weekly, 75% use AI for half or more of their work, and 56% report doing 70%+ of their work with AI. Among readers of The Pragmatic Engineer, AI is now mainstream in day-to-day software engineering work.

#7
SNS Insider 2025-12-01 | AI Code Tools Market Size, Share & Growth Report 2035 - SNS Insider
NEUTRAL

The AI Code Tools Market Size was valued at USD 7.59 billion in 2025. By Offering: In 2025, Tools led the market with share 76.40%. These tools allow developers to work faster with shorter development cycles and achieve better coding quality.

#8
ShiftMag 2025-12 | 42% of Code Is Now AI-Assisted! - ShiftMag
SUPPORT

Developers estimate that 42% of the code they commit is AI-assisted... That number is expected to rise to 65% by 2027... Sonar surveyed over 1,100 developers worldwide.

#9
Futurum Group 2026-01-01 | AI Reaches 97% of Software Development Organizations
NEUTRAL

76.6% of organizations actively using AI in development workflows: 38.7% in some projects, 21.1% extensively across most projects, and 16.8% piloting in limited use. This 97% adoption trajectory validates that 2026 marks the inflection point where developers become engineers of agent-driven development.

#10
Greptile 2025-11-01 | The State of AI Coding 2025 | Greptile
NEUTRAL

Lines of code per developer grew from 4,450 to 7,839 as AI coding tools act as a force multiplier. AI Memory Packages: mem0 dominates with 59% market share.

#11
Itransition 2026-01-01 | Software Development Statistics for 2026: Key Facts & Trends
NEUTRAL

85% of developers regularly use AI tools for writing code and development, and 62% rely on at least one AI coding assistant in their workflows (JetBrains). 84% of professionals are using or planning to use AI tools in their software development process (Stack Overflow).

#12
Intel Market Research 2025-09-26 | AIPowered Coding Assistant Market Outlook 2025-2032
REFUTE

The global AI-powered coding assistant market was valued at USD 205 million in 2024 and is projected to grow from USD 220 million in 2025 to USD 341 million by 2032, exhibiting a CAGR of 7.0% during the forecast period.

#13
Master of Code 2026-01-01 | 350+ Generative AI Statistics [January 2026] - Master of Code
NEUTRAL

According to Gartner, 40% of enterprise applications will include task-specific AI agents by the end of 2026, up from less than 5% just a year earlier. However, 68% of institutions have moved fewer than 30% of their artificial intelligence experiments into full production.

#14
Panto AI 2026-01 | AI Coding — Key Statistics & Trends (2026) - Panto AI
REFUTE

AI adoption: 91% across their sample of 135,000+ developers... 22% of merged code was AI-authored... Share of merged code AI-authored (DX sample): 22%.

#15
LLM Background Knowledge 2026-03 | Consensus on AI Code Generation Trends
REFUTE

Major surveys from GitHub, Stack Overflow, and JetBrains in 2025 consistently report AI assistance in 20-40% of code, but human developers write, review, and modify the majority; no primary data supports AI generating over 50% as of 2026.

#16
CodeRabbit 2026-01-01 | 2025 was the year of AI speed. 2026 will be the year of AI quality.
NEUTRAL

The year 2025 will be remembered as the moment AI-assisted software development entered its acceleration era.

#17
Netcorp Software Development 2026-01 | AI-Generated Code Statistics 2026: Can AI Replace Your ... - Netcorp
SUPPORT

According to recent global estimates, 41% of all code is now AI-generated... As of early 2025, 25% of Google’s code was AI-assisted, but according to Google CEO Sundar Pichai... the real focus is on engineering velocity.

#18
Mocha 2026-01 | AI App Builder Statistics 2026: 50+ Key Data Points - Mocha
NEUTRAL

30%+ of all new code at Google is AI-generated, up from 25% six months prior (Sundar Pichai, Alphabet Q1 2025 earnings call)... 41% of all code written globally is AI-generated (JetBrains Developer Ecosystem 2025)... GitHub Copilot generates an average of 46% of code for its users.

#19
EliteBrains 2025-12 | AI-Generated Code Stats 2026: How Much Is Written by AI?
SUPPORT

Here's the shocking truth: 41% of code is now AI-generated... As of 2024, 256 billion lines of code have already been generated by AI.

#20
Tianpan Forum 2026-01 | 41% of All Code in 2026 Is AI-Generated. Google Says 25%. Which ...
NEUTRAL

Industry surveys report 41% of all code is now AI-generated; Google publicly states 25% of their code is AI-generated (Sundar Pichai, Oct 2024).

Full Analysis

Expert review

How each expert evaluated the evidence and arguments

Expert 1 — The Logic Examiner

Focus: Inferential Soundness & Fallacies
False
2/10

Across the evidence pool, the only quantified measures of AI-authored code cluster well below a majority—e.g., 29% AI-written in a large GitHub Python contribution analysis (Source 3), 22% of merged code AI-authored in a large developer sample (Source 14), and JetBrains' estimate of ~30–40% AI-assisted/lines touched (Source 2), while adoption/"essential" reliance stats (Sources 1, 4, 9, 11) do not logically entail that AI generates >50% of all code. The proponent's inference from high usage and selected high-intensity/self-selected surveys (Source 6) to a global majority commits scope/representativeness errors and a threshold leap, so the claim that AI is responsible for generating the majority of software code as of 2026 is not supported and is more likely false given the available quantified estimates.

Logical fallacies

Hasty generalization / selection bias: extrapolating from a self-selected newsletter audience with unusually high AI usage (Source 6) to the global share of all code.Scope mismatch: treating "AI-assisted" or "lines touched" (Source 2) and "use AI for half or more of their work" (Source 6) as equivalent to "majority of all software code generated by AI".Non sequitur / threshold leap: inferring that figures in the ~30–40% range imply the global aggregate has crossed >50% without evidence of the required distribution/weighting by code volume.Equivocation: sliding between "AI-assisted," "AI-authored drafts later edited," and "AI-generated" as if they are the same quantity for purposes of the majority claim.
Confidence: 8/10

Expert 2 — The Context Analyst

Focus: Completeness & Framing
False
2/10

The claim frames high adoption and “AI-assisted/AI-touched” workflow metrics as if they were equivalent to AI authoring most code, but the evidence distinguishes assistance from authorship and reports materially lower AI-authored shares in real-world code artifacts (e.g., ~30–40% AI-assisted/lines touched in JetBrains [2], 29% AI-written in a large GitHub Python contribution analysis [3], and 22% AI-authored merged code in a large developer sample [14]). With that context restored, there is no solid basis to conclude AI is generating >50% of all software code as of 2026, so the overall impression is false.

Missing context

Key definitional ambiguity: “responsible for generating” could mean AI-authored, AI-assisted, AI-suggested, or AI-edited; several cited figures are explicitly about assistance/lines touched rather than authorship (JetBrains [2]).Representativeness: pro evidence leans on a highly self-selected audience survey (The Pragmatic Engineer [6]) and on secondary compilations that may misquote primary sources (e.g., Mocha/Netcorp/EliteBrains [18][17][19]) rather than global, cross-language measurement.Scope mismatch: some empirical measurements are narrow (Python on GitHub, merged code in a specific pipeline) and may not cover all software code, but they still undercut the “majority” framing absent any broader >50% primary measurement (Trending Topics [3], Panto AI [14], synthesis [15]).
Confidence: 8/10

Expert 3 — The Source Auditor

Focus: Source Reliability & Independence
False
2/10

The most authoritative and methodologically grounded sources consistently refute the "majority" claim: Source 3 (Trending Topics, citing a peer-reviewed-style analysis of 30 million real GitHub Python contributions) places AI-written code at 29% by end-2025; Source 14 (Panto AI, drawing on a DX sample of 135,000+ developers) finds only 22% of merged code is AI-authored; Source 2 (JetBrains, 24,000+ developer survey) explicitly clarifies that the 41% figure refers to lines "touched" by AI, not independently generated; and Source 15 (LLM Background Knowledge synthesizing GitHub, Stack Overflow, and JetBrains surveys) explicitly states no primary data supports AI generating over 50% as of 2026. The strongest supporting source, Source 6 (The Pragmatic Engineer), is self-admittedly drawn from a newsletter audience self-selected for AI enthusiasm, making it wholly unrepresentative of the global developer population, while Sources 17–20 are low-authority blogs and forums that largely recycle the same JetBrains "41% touched" figure without independent verification. The claim that AI is responsible for generating the majority of software code as of 2026 is clearly false based on what the most reliable, independently gathered evidence actually shows.

Weakest sources

Source 6 (The Pragmatic Engineer) is unreliable for this claim because it surveys readers of an AI-focused newsletter, a population self-selected for high AI tool adoption, making it wholly unrepresentative of the global developer population.Source 17 (Netcorp Software Development) is a low-authority blog that misrepresents the JetBrains '41%' figure as 'AI-generated' when JetBrains explicitly clarifies it means lines 'touched' by AI, not independently generated.Source 19 (EliteBrains) is a low-authority blog that repeats the same mischaracterized '41% AI-generated' statistic without independent verification or methodological transparency.Source 20 (Tianpan Forum) is a user forum post with no editorial oversight or independent research, making it unsuitable as evidence for a quantitative claim.Source 18 (Mocha) is a low-authority blog that aggregates and conflates figures from disparate sources (GitHub Copilot user averages, Google executive statements, JetBrains survey data) without accounting for their differing scopes and methodologies.
Confidence: 7/10

Expert summary

See the full panel summary

Create a free account to read the complete analysis.

Sign up free
The claim is
False
2/10
Confidence: 8/10 Unanimous

The arguments

Two AI advocates debated this claim using the research gathered.

Argument for

P
Proponent Argues TRUE

By 2026, multiple independent indicators show AI is already producing at least half of developers' output: The Pragmatic Engineer's 2026 survey reports 75% of respondents use AI for half or more of their work and 56% use AI for 70%+ (Source 6, The Pragmatic Engineer), while GitHub Copilot is reported to generate ~46% of code for its users and Google reports 30%+ of new code is AI-generated—evidence of AI frequently being the primary code producer in real-world pipelines (Source 18, Mocha). Given near-ubiquitous adoption (85%+ regular use and “essential workflow” reliance for a majority) and documented AI-authorship shares already approaching the 40–50% range (Sources 2 and 4, JetBrains; Source 8, ShiftMag), the most coherent reading is that the global aggregate crossed the “majority of code written” threshold by 2026 as high-intensity AI users dominate net new code volume.

O
Opponent Rebuttal

Your reliance on Source 6 (The Pragmatic Engineer) is fatally undermined by its own admission of severe selection bias — it surveys readers of a newsletter self-selected for AI enthusiasm, making it wholly unrepresentative of the global developer population, and you cannot extrapolate from that skewed sample to a worldwide majority claim. Furthermore, your logical leap that "approaching 40–50%" equals a crossed majority threshold is precisely that — a leap, not evidence — and it is directly demolished by Source 14's finding of only 22% AI-authored merged code across 135,000+ real developers, Source 3's empirical analysis of 30 million actual GitHub contributions placing AI-written code at just 29%, and Source 15's explicit synthesis concluding that no primary data supports AI generating over 50% as of 2026.

Argument against

O
Opponent Argues FALSE

The claim that AI generates the "majority" of software code is directly contradicted by the most rigorous and authoritative evidence available: Source 3 places AI-written code at only 29% of contributions as of end-2025 based on analysis of 30 million real GitHub contributions, Source 14 finds just 22% of merged code is AI-authored across 135,000+ developers, and Source 15 — synthesizing major surveys from GitHub, Stack Overflow, and JetBrains — explicitly states that "no primary data supports AI generating over 50% as of 2026." Even the figures most favorable to the claim, such as the 41% cited in Source 2 (JetBrains), explicitly clarify that this refers to lines "touched" by AI, not fully and independently generated, meaning the actual majority of code remains human-written, reviewed, and controlled.

P
Proponent Rebuttal

You treat Source 3 (Trending Topics) and Source 14 (Panto AI) as if they measure “all software code,” but they're narrow operationalizations (Python GitHub contributions; “merged code” in a particular DX sample) that can't validly refute a global, cross-language, net-new-code claim—especially when broader workflow surveys show many developers using AI for half or more of their work (Source 6, The Pragmatic Engineer) and widespread “essential” reliance (Source 4, JetBrains). And your appeal to Source 15 (LLM Background Knowledge) is an argument from authority that doesn't outweigh primary survey evidence of high-intensity AI use, while your “lines touched ≠ generated” point (Source 2, JetBrains) is a definitional dodge that ignores that “responsible for generating” includes AI-authored drafts later edited by humans rather than only fully autonomous code.

Your annotation will be reviewed by an editor before becoming visible.

Embed this verification

Copy this code and paste it in your article's HTML.