Around one-third of newly written computer code in the US was supported by artificial intelligence by the end of 2024, according to an analysis of GitHub data. While AI tools are widely used by less experienced programmers, productivity gains were found to benefit mainly seasoned developers, potentially widening existing skill gaps
Representational Image
By the end of 2024, around one-third of newly written blocks of computer programs in the US took support from AI systems -- gains in productivity, however, went to seasoned programmers, according to an analysis of patterns seen on a widely used collaborative coding platform.
AI-supported coding was found to be high also in other countries, including Germany and France, with India catching up fast.
"We analysed more than 30 million Python (programming language) contributions from roughly 160,000 developers on GitHub, the world's largest collaborative programming platform," author Simone Daniotti, from Austria's Complexity Science Hub and Utrecht University in the Netherlands, said.
Trained on vast amounts of text data, generative AI can respond to a user's requests in the natural language, including writing and debugging computer code.
Findings published in the journal Science also show that while AI usage is highest among less experienced programmers, productivity gains go to seasoned developers -- generative AI can therefore widen existing gaps, instead of levelling the playing field.
Researchers used a specially trained AI model to identify whether blocks of computer code were AI-generated, for instance via ChatGPT or GitHub Copilot.
The team was able to track programming work across the globe in real time as GitHub records every step of coding -- additions, edits, improvements. Python is among the most widely used programming languages in the world.
"In the US, AI-assisted coding jumped from around five per cent in 2022 to nearly 30 per cent in the last quarter of 2024," author Frank Neffke, who leads the transforming economies group at Complexity Science Hub, said.
"While the share of AI-supported code is highest in the US at 29 per cent, Germany reaches 23 per cent and France 24 per cent, followed by India at 20 per cent, which has been catching up fast," he said.
Neffke added that Russia (15 per cent) and China (12 per cent) still lagged behind at the end of the study.
"Currently AI writes an estimated 29 per cent of Python functions in the US, a shrinking lead over other countries," the authors wrote.
The study also showed that the use of generative AI increased programmers' productivity by 3.6 per cent by the end of 2024.
"That may sound modest, but at the scale of the global software industry it represents a sizeable gain," Neffke said.
Less experienced programmers were seen to use generative AI in 37 per cent of their code, compared to 27 per cent for experienced programmers.
However, productivity gains are driven exclusively by experienced users, the researchers found.
"Beginners hardly benefit at all," Daniotti said, adding that generative AI therefore does not automatically level the playing field -- it can widen existing gaps.
Experienced software developers were also seen to experiment more with new libraries and unusual combinations of existing software tools.
"This suggests that AI does not only accelerate routine tasks, but also speeds up learning, helping experienced programmers widen their capabilities and more easily venture into new domains of software development," author Johannes Wachs, a faculty member at Complexity Science Hub, said.
This story has been sourced from a third party syndicated feed, agencies. Mid-day accepts no responsibility or liability for its dependability, trustworthiness, reliability and data of the text. Mid-day management/mid-day.com reserves the sole right to alter, delete or remove (without notice) the content in its absolute discretion for any reason whatsoever.
Subscribe today by clicking the link and stay updated with the latest news!" Click here!
