By Isabelle Bousquette

After years of hype and panic over how many jobs AI will extinguish or replace, it's clear the technology is bringing some big changes to the coding workforce.

AI coding tools, which can automate large portions of code development, emerged as an early generative AI use case, driving double digit percentage efficiency gains and helping more code get written faster. Microsoft-owned GitHub Copilot, one of the more popular coding tools, has been adopted by more than 77,000 organizations in the roughly two years since its release, Microsoft said in its fourth-quarter earnings.

The tools are designed to supplement rather than replace human workers and their output still needs to be double-checked, companies say. But their growing adoption is quickly reshaping the size and scope of the development teams.

"2025 is going to be a very fascinating year with some of these tools, as we start to scale," said KeyBank Chief Information Officer Amy Brady. "We're not far enough on the journey where I can confidently say it's going to replace all entry-level code generation. Do I think it could replace some? Yes."

Companies using the tools say they can get the same amount of work or more done with leaner teams. Automating the mundane boilerplate coding lets developers focus on complex problem solving and skills associated with using the AI tools, they say. And when hiring developers, companies say they can be more tempered and selective than several years ago when the talent market was so tight they were desperate to get bodies into seats.

"It really changes and morphs the way you think about a traditional developer," said Yang Lu, the CIO of Coach and Kate Spade owner Tapestry about GenAI coding tools. The fashion company is seeing roughly 10% to 20% efficiency gains from its use of GitHub Copilot.

"We can get through more [coding] quicker and do more tests and learn quicker," Lu said. "Like, how much time do you spend debugging because you put the bracket in the wrong place? That's the sort of time spent that's unnecessary," she said.

For the developers, priorities are shifting more toward how to use and prompt the AI tools, she said. "We're training developers to think more from a prompting perspective versus the traditional programming perspective," she said. "It's a different mindset."

Silicon Valley is not immune to the shift. Alphabet Chief Executive Sundar Pichai recently said more than a quarter of the company's new code is AI-generated.

While there are still skills shortages in some areas, those companies that are hiring developers are finding an easier market than they did a couple of years ago, said Ryan Sutton, technology hiring and consulting expert at recruiting firm Robert Half. He added, though, that the more employer-friendly market is attributable to several factors, including the economy and a normalization after the pandemic-fueled hiring boom.

Companies that might have compromised on what they were looking for in the past don't anymore, Sutton said, although some of the skill sets have changed. Companies want candidates who can go beyond writing lines of code and think critically about how to solve problems through technology. Communication skills are also desired, he added.

"The world is becoming more competitive. The bar for talent and the expectation of talent has just risen," said Jason Gowans, Levi Strauss & Co. chief digital and technology officer. Levi's last year rolled out GitHub Copilot and quickly found that users were accepting the coding suggestions 40% of the time after just the first prompt.

"It is possible that we might see a future where some workers are displaced," Gowans said, adding that top talent will always remain in demand.

The unemployment rate in the information technology sector rose from 3.9% in December to 5.7% in January, well above the month's overall jobless rate of 4%, according to a report from consulting firm Janco Associates based on data from the U.S. Department of Labor. Victor Janulaitis, CEO of Janco, attributes part of the shift to the growing use of AI.

TD Bank Group continues to hire developers, and Brent Foster, vice president of software and quality engineering practices at the financial services firm, said he's looking for candidates who can focus on the things the tools don't do themselves. These include prompt engineering and working at a higher level of abstraction, stitching together different building blocks of existing services to build an architecture.

TD Bank Group said 75% of its engineers who were using GitHub Copilot felt like they were being equally or more productive than before the bank started rolling it out. "What it has done is really emphasize the clarity we have around being focused on skills and being focused on outcomes and being focused on bringing the best talent in," Foster said.

To be sure, these coding generation tools remain in their infancy and CIOs say the gap between what they can do today and their future potential is big.

At Verizon, Chief Data and AI Officer Kalyani Sekar said the tools are helping the team work through a backlog of requests from the business at a faster pace -- but not so much that she's ready to think about staffing changes of either internal employees or external contractors.

"Maybe when the backlog is going to a very nominal level. And at that time we really need to think about it," she said. But she added it's not clear exactly when that will be.

"I won't claim victory at this point in time."

Write to Isabelle Bousquette at isabelle.bousquette@wsj.com

(END) Dow Jones Newswires

March 04, 2025 07:00 ET (12:00 GMT)

Copyright (c) 2025 Dow Jones & Company, Inc.