Software engineering was considered the easiest victory of artificial intelligence. Today companies like OpenAI, Anthropic, Microsoft, and Google have released AI products specifically designed for coding. And a survey of nearly 5,000 technology professionals released in a report last year by Google’s DevOps Research and Assessment (DORA) team found that 90 percent of respondents said they were using AI at work – with more than 80 percent saying the technology had increased their productivity.
“We see a huge number of people who are relying on AI to get their work done, at least to a moderate amount, which is really fascinating,” says Nathan Harvey, who leads the Dora team.
AI can generate code for everything from web and mobile apps to data management tools. It often automates some of the more difficult elements of the work, such as building testing infrastructure and updating software to run on new devices and systems. In some cases, even inexperienced developers can create working prototypes by describing their intentions for an AI system, often referred to as “vibe coding,” a term coined by OpenAI co-founder and researcher Andrzej Karpathy. But writing code is only part of the job; Developers still have to verify that it’s doing what it’s supposed to do and fix it if it fails.
On supporting science journalism
If you enjoyed this article, consider supporting our award-winning journalism Subscribing By purchasing a subscription, you are helping ensure a future of impactful stories about the discoveries and ideas shaping our world today.
Another finding of the DORA report was that while individual coder effectiveness increased with the use of AI, so too did “software delivery instability” – a measure of how often code needed to be reverted or patched after release to address unexpected issues.
“As you use more AI, you’re more likely to roll back changes you make in production,” says Harvey. “And that, obviously, is something you want to avoid.”
Even though it is becoming increasingly efficient at writing code, AI does not eliminate the need for human software engineering. Developers often still need to craft custom code to handle unusual cases or specific business needs – or at least tweak the AI tool’s output that may not be reflected in the AI training data. They still need to carefully confirm that machine-generated programs behave exactly as intended and meet company standards.
AI tools don’t automatically shorten the workday. Studies show that in some workplaces, AI has increased the pressure to move faster than ever before.
If employers don’t manage its effects, AI could also increase stress and burnout among software engineers. In a report published in Harvard Business Review In February, researchers at the Haas School of Business at the University of California, Berkeley found that employees at a US tech company completed more tasks, worked faster and worked more hours after adopting AI. Even without the company mandating the use of the technology, employees began prompting AI during lunches, breaks and meetings, with some finding the former downtime less refreshing. Researchers warn that there is a risk that the initial enthusiasm and productivity boost may lead to burnout, lower quality output and higher employee turnover.
This pressure is not happening in a vacuum. After years of industry-wide layoffs and corporate mandates for efficiency, AI is often deployed with the expectation that those left behind will do more with less.
Additionally, a report assessing more than 500 developers released late last year by Multitudes, a New Zealand-based company that helps businesses track and optimize software engineering practices, indicated that AI could increase workers’ productivity as well as work hours. On average, engineers merged 27.2 percent more “pull requests” – packages of code that were approved for inclusion in existing software projects. But they also experienced a 19.6 percent increase in “out-of-hours commitments” – submitting coding work outside their normal schedule. This may be a sign of problems to come.
“If work is increasing outside of hours, it’s not good for the individual,” says Lauren Peet, founder and CEO of Multitudes. “This can lead to burnout.”
The Multitudes report certainly doesn’t prove that AI has directly caused measured changes, but Peet says the interview suggests that the observed change in hours worked among engineers is likely a sign that businesses expect more productivity from employees in the AI age.
“People were feeling extra pressure to work more, and that seemed to be contributing to them putting in more hours,” she says.
While some research has suggested that less experienced developers may be among those who benefit most from the assistance of AI, and Vibe coding may allow people with minimal programming background to create running programs, a recent evaluation from Anthropic suggests that over-reliance on AI may impact the development of coding skills.
In a report released in January, Anthropic researchers found that software engineers working with a new software library saw a small, statistically insignificant increase in speed when solving a task with the assistance of AI, compared to a control group working without AI assistance. However, when coders were quizzed about the software library after the task, the group given AI assistance scored 17 percent lower than the AI-free group. Those who asked questions about AI rather than relying on AI to generate code generally performed better, but researchers raised concerns that using AI to complete tasks as quickly as possible under workplace pressure could be detrimental to engineers’ professional development.
Additionally, they noted, the biggest differences in quiz performance were in questions related to debugging code – the process of finding and fixing flaws in code that cause malfunctions. In other words, junior developers who rely heavily on AI may have difficulty not only writing the code themselves, but also understanding and finalizing the code generated in the first place. in a statement to scientific American, Anthropologist researcher Judy Hanwen Shen said that the goal “should not be to use AI to avoid cognitive effort – it should be to use AI to deepen it.”
Already, UC Berkeley researchers note, engineers may find themselves helping colleagues who have created incomplete software solutions through Vibe coding. And some open-source projects have reported an increase in low-quality, AI-driven submissions that waste lead developers’ time.
This comes after a 2025 Harvard Business School working paper indicated that AI could shift open-source developers’ time from handling project management tasks, such as reviewing code contributions and maintaining lists of issues for contributors to fix, to generating code themselves.
“Now you can do it yourself, so there’s not as much need to interact with others,” says Manuel Hoffman, co-author of the paper and assistant professor of information systems at the Paul Maraz School of Business at the University of California, Irvine. “And that’s not necessarily a bad thing.”
Nevertheless, such use of AI may limit another channel for less experienced programmers to hone their skills, develop professional networks, and expand their resumes.
And as AI redefines the meaning of productivity, workplace structures that prevent burnout, keep workloads manageable, and provide pathways for advancement and training may be more important than ever.
“When you have great things going on, and you add some AI to the mix, they’ll probably get even better,” says Harvey. “And when painful things are happening to you, [and] You add some AI to the mix, [you’re] “Maybe that pain will be felt a little more intensely.”
<a href