During a recent internal discussion at Ballast Lane Applications, one of our engineers mentioned something that caught everyone's attention: "With AI, you don't stop coding, and time just goes on." Others immediately agreed. They even called their relationship with AI coding tools "addictive."
It wasn’t a complaint; in fact, it was a reflection on how AI tools have shifted their coding experience.
Artificial intelligence promised to make software engineering easier with faster code generation, automated testing, and intelligent debugging. In many ways, it has delivered on that promise. However, this progress has also created some unexpected patterns worth paying attention to.
A recent study from UC Berkeley's Haas School of Business tracked knowledge workers using AI tools and found something curious: people were working longer hours, taking fewer breaks, and reporting higher engagement with their work, even during the time they'd normally spend disconnected.
At the end of the day, we aren't just looking at a shift in hours logged. AI tools are changing how we actually experience our day-to-day jobs.
AI Changed How Work Feels
The UC Berkeley researchers found that workers using AI tools took on responsibilities that hadn't previously been theirs. Not because anyone asked them to, but because AI made those tasks suddenly feel accessible.
When you can generate a data visualization in minutes instead of hours, you start offering to create them during meetings. When you can draft a complex function with a few prompts, you experiment during lunch breaks. When you can solve problems that once required help from other teams, you do it yourself.
The pattern isn't about AI creating more work. It's about AI making work feel more engaging and accessible, so much so that the traditional boundaries between work time and personal time start to blur naturally.
In tech companies, engineers describe prompting AI tools during breaks, running multiple assistants simultaneously, and "experimenting" during evenings in ways that feel creative rather than burdensome.
The researchers noted something interesting: workers couldn't always distinguish between "using AI to work" and "working." The tool and the task had merged in a way that felt different from traditional software development.
Why AI Coding Feels Particularly Engaging
Every software engineer knows the satisfaction of solving a difficult problem. The moment the code compiles successfully. The relief when tests pass. These small wins trigger dopamine release, a signal from the brain's reward system that reinforces behavior.
A dopamine positive feedback loop occurs when a behavior is followed by a reward; you're highly likely to repeat that action. AI tools have accidentally created a feedback loop that makes it psychologically difficult to stop working. AI suggestions create an unpredictable reward schedule.
Sebastian Beltran, a Ballast Lane Applications front-end developer, said:
“AI shrinks the gap between having an idea and seeing it work to basically a prompt. You finish something, get that rush of 'wow, it actually did it,' notice something else you could improve, and go 'just one more prompt.' Next thing you know, it's 2 AM, and you've been at it for six hours straight.”
When "Free Time" Becomes "Productive Time"
The UC Berkeley researchers found that workers using AI tools reported thinking about work during non-work hours 40% more frequently than those without access to AI. But here's the catch: they weren't necessarily stressed about work. They were engaged with it.
Engineers describe "experimenting" with AI tools on weekends. That experimentation often translates into production code, feature proposals, and technical decisions. The energy spent, the problems solved, and the code written- it's all real work happening outside traditional work hours. But it doesn't always feel like work in the conventional sense.
This creates an interesting challenge: How do you set boundaries around something that feels rewarding rather than exhausting?
The Path Forward: Using AI Consciously
AI tools have genuinely made software engineering more efficient and more engaging, and that's worth celebrating. But they've also blurred boundaries that used to be clearer between work and rest, creation, professional time, and personal time. That's worth paying attention to.
The UC Berkeley research suggests that AI doesn't reduce work—it intensifies it. Not necessarily in a bad way, but in a way that makes it harder to naturally disengage.
The engineers who succeed won’t be the ones who use AI the most, but the ones who learn to use these tools with awareness of the behaviors they create, and with intentional practices to maintain balance. That means noticing when "just one more prompt" at 10 PM has become a nightly habit or recognizing when weekend "experiments" are actually unpaid work.
At Ballast Lane Applications, we're committed to helping our teams navigate this. Not by restricting tool access or creating rigid policies, but by creating awareness, sharing what we're learning, and creating space for sustainable practices. We've partnered with Selia, a mental health platform that connects our team with over 500 certified psychologists who understand their culture, context, and language. It's accessible, confidential, and free for all team members because mental health support shouldn't be a barrier. If you want to read more about our well-being strategy in detail, see Engineering Excellence in the AI Era: Maximizing Output and Well-being.