Who is using AI to code? Global diffusion and impact of generative AI

Simone Daniotti; Johannes Wachs; Xiangnan Feng; Frank Neffke

doi:10.1126/science.adz9311

Who is using AI to code? Global diffusion and impact of generative AI

Simone Daniotti^*
, Johannes Wachs
, Xiangnan Feng
, Frank Neffke

^*Corresponding author for this work

Innovation Studies

Research output: Contribution to journal › Article › Academic › peer-review

Abstract

Generative coding tools promise big productivity gains, but uneven uptake could widen skill and income gaps. We train a neural classifier to spot artificial intelligence (aI)–generated Python functions in more than 30 million GitHub commits by 160,097 software developers, tracking how fast, and where, these tools take hold. Currently, aI writes an estimated 29% of Python functions in the uS—a shrinking lead over other countries. We estimate that quarterly output, measured in online code contributions, consequently increased by 3.6%. aI seems to benefit experienced, senior-level developers: They increased productivity and more readily expanded into new domains of software development. by contrast, early-career developers showed no significant benefits from aI adoption. This may widen skill gaps and reshape future career ladders in software development.

Original language	English
Pages (from-to)	831-835
Number of pages	5
Journal	Science
Volume	391
Issue number	6787
DOIs	https://doi.org/10.1126/science.adz9311
Publication status	Published - 19 Feb 2026

Bibliographical note

Publisher Copyright:
Copyright © 2026 the authors, some rights reserved;

Access to Document

10.1126/science.adz9311

Cite this

@article{4450b18f9c584132be3418858fc1387c,

title = "Who is using AI to code? Global diffusion and impact of generative AI",

abstract = "Generative coding tools promise big productivity gains, but uneven uptake could widen skill and income gaps. We train a neural classifier to spot artificial intelligence (aI)–generated Python functions in more than 30 million GitHub commits by 160,097 software developers, tracking how fast, and where, these tools take hold. Currently, aI writes an estimated 29\% of Python functions in the uS—a shrinking lead over other countries. We estimate that quarterly output, measured in online code contributions, consequently increased by 3.6\%. aI seems to benefit experienced, senior-level developers: They increased productivity and more readily expanded into new domains of software development. by contrast, early-career developers showed no significant benefits from aI adoption. This may widen skill gaps and reshape future career ladders in software development.",

author = "Simone Daniotti and Johannes Wachs and Xiangnan Feng and Frank Neffke",

note = "Publisher Copyright: Copyright {\textcopyright} 2026 the authors, some rights reserved;",

year = "2026",

month = feb,

day = "19",

doi = "10.1126/science.adz9311",

language = "English",

volume = "391",

pages = "831--835",

journal = "Science",

issn = "0036-8075",

publisher = "American Association for the Advancement of Science",

number = "6787",

}

TY - JOUR

T1 - Who is using AI to code? Global diffusion and impact of generative AI

AU - Daniotti, Simone

AU - Wachs, Johannes

AU - Feng, Xiangnan

AU - Neffke, Frank

PY - 2026/2/19

Y1 - 2026/2/19

N2 - Generative coding tools promise big productivity gains, but uneven uptake could widen skill and income gaps. We train a neural classifier to spot artificial intelligence (aI)–generated Python functions in more than 30 million GitHub commits by 160,097 software developers, tracking how fast, and where, these tools take hold. Currently, aI writes an estimated 29% of Python functions in the uS—a shrinking lead over other countries. We estimate that quarterly output, measured in online code contributions, consequently increased by 3.6%. aI seems to benefit experienced, senior-level developers: They increased productivity and more readily expanded into new domains of software development. by contrast, early-career developers showed no significant benefits from aI adoption. This may widen skill gaps and reshape future career ladders in software development.

AB - Generative coding tools promise big productivity gains, but uneven uptake could widen skill and income gaps. We train a neural classifier to spot artificial intelligence (aI)–generated Python functions in more than 30 million GitHub commits by 160,097 software developers, tracking how fast, and where, these tools take hold. Currently, aI writes an estimated 29% of Python functions in the uS—a shrinking lead over other countries. We estimate that quarterly output, measured in online code contributions, consequently increased by 3.6%. aI seems to benefit experienced, senior-level developers: They increased productivity and more readily expanded into new domains of software development. by contrast, early-career developers showed no significant benefits from aI adoption. This may widen skill gaps and reshape future career ladders in software development.

UR - https://www.scopus.com/pages/publications/105031597198

U2 - 10.1126/science.adz9311

DO - 10.1126/science.adz9311

M3 - Article

C2 - 41570112

AN - SCOPUS:105031597198

SN - 0036-8075

VL - 391

SP - 831

EP - 835

JO - Science

JF - Science

IS - 6787

ER -

Who is using AI to code? Global diffusion and impact of generative AI

Abstract

Bibliographical note

Access to Document

Other files and links

Embargoed Document

Fingerprint

Cite this