Artificial intelligence (AI) was expected to revolutionize software engineering, and to some extent, it has. Nearly 90% of tech professionals now use AI tools at work, with over 80% reporting productivity gains. Yet, despite the hype, data shows a disturbing trend: developers are working longer hours, not less. The promise of AI automating tedious tasks and boosting efficiency is colliding with the reality of unstable code, increased pressure, and potential burnout.
The Paradox of Productivity
AI can generate code for web apps, mobile software, and data tools – even allowing inexperienced developers to create basic prototypes through “vibe coding.” However, AI-generated code is rarely flawless. Developers still spend significant time verifying outputs and patching errors, leading to a rise in “software delivery instability.” The DORA report shows that more AI use correlates with more frequent rollbacks and fixes. This means that while individual coding speed may increase, the overall process can become more fragile.
Pressure to Perform
The issue isn’t just technical; it’s also cultural. AI is often deployed alongside expectations of increased output with fewer resources. Companies expect more from employees in the AI era, leading to pressure to work faster, even during off-hours. Studies from Berkeley show that employees adopting AI took on more tasks, worked at a quicker pace, and logged more hours. Many now prompt AI during breaks and meetings, blurring the lines between work and personal time.
The Rise in Out-of-Hours Work
Multitudes reports that engineers are merging 27.2% more pull requests, but also submitting 19.6% more code outside of normal work hours. This isn’t just a matter of improved efficiency; it suggests employers are pushing for greater productivity, potentially leading to burnout. Lauren Peate, CEO of Multitudes, warns that this trend is “not good for the person.”
The Skill Gap
Overreliance on AI can also hinder skill development. Anthropic research found that engineers who heavily rely on AI scored 17% lower on coding knowledge tests compared to those who didn’t. The biggest gap was in debugging – the ability to find and fix flaws in code. Using AI as a shortcut may leave junior developers unable to understand or refine the AI-generated output, potentially worsening the long-term quality of work.
The Future of Software Engineering
The pressure isn’t just affecting individual developers; it’s shifting how open-source projects function. There’s a reported rise in low-quality, AI-driven submissions that consume core developers’ time, and a decline in collaborative project management.
Ultimately, AI doesn’t eliminate the need for human expertise; it reshapes it. The key question is whether workplaces will adapt to prevent burnout, manage workloads, and provide training opportunities, or whether the promise of AI will simply translate into longer hours and greater strain on software engineers.
The reality is that AI amplifies existing dynamics: it makes good things better, but also makes bad things worse. The challenge is not just about using the tools, but about building a sustainable ecosystem around them.
