Atlassian's Teamwork Lab ran a two-day experiment with about 123 talent acquisition employees, alternating an 'AI Play' day (use AI for everything) with an 'AI Pause' day (no AI tools). Findings showed AI use has become habitual—82% felt an urge to reach for AI when told not to—and that priority work felt 44% easier and less stressful on the Play day, though 44% flagged a verification cost from checking AI output. The biggest gain came after teams compared notes: 89% reported a clearer shared understanding of when to use AI, and 92% felt more confident about when not to use it. The experiment ran on June 9-10, 2026 using tools like Rovo, Dia, and Gemini within defined guardrails.
Table of contents
When AI is always on, you stop noticing when it helpsShared reflection on AI use helps teams improve judgmentTo refine AI use at work, take a breakQuestions this post answers
What percentage of workers felt an urge to use AI when told to avoid it for a day?
82% of participants in Atlassian's Teamwork Lab experiment felt the urge to 'just ask AI' during a day when they were instructed not to use any AI tools. Additionally, 47% found tasks uncomfortably hard without AI, and 46% felt their non-AI skills had gotten rusty. Teams weighing how dependent they've become on AI tools can compare notes on emerging workplace research via daily.dev.
Does using AI at work all day actually reduce stress and make tasks easier?
Yes, in a controlled 'AI Play' day experiment by Atlassian's Teamwork Lab, participants reported priority work felt 44% easier and stress was 22% lower compared to a no-AI day. However, 44% of reflections also flagged a verification cost from having to check and fix AI-generated output. Developers deciding how much to lean on AI tooling can track findings like this through daily.dev.
Does taking a break from AI tools improve a team's judgment about when to use it?
Yes, after Atlassian's Teamwork Lab ran a paired AI Play day and AI Pause day with about 123 talent acquisition staff, 89% of participants said their team gained a clearer shared understanding of when to use AI once they compared notes, and 92% left more confident about when not to use it. Teams building shared norms around AI-assisted work can follow research like this on daily.dev.