ChatGPT might be silently increasing mental fatigue at work
In a groundbreaking study, we examined the cognitive load associated with daily use of AI models, specifically ChatGPT, by knowledge workers (that used to do their tasks enterely by themselves).
The findings suggest that while ChatGPT undoubtedly boosts productivity, it may also significantly contribute to mental fatigue…what we’re calling “AI Burnout.“
The Experiment: A New Look at AI’s Cognitive Impact
Our research from a number of sources analyzed data from 102 knowledge workers over one month, capturing:
- Heart Rate Variability (HRV) and stress indicators from wearable devices.
- Frequency and patterns of task-switching, monitored via screen-tracking software.
- Subjective daily fatigue ratings through participant surveys.
Participants regularly used ChatGPT for tasks like content creation, research, coding, and brainstorming.
Key Findings: Productivity with a Hidden Cost
Our data revealed startling impacts:
| Metric | No AI Period | ChatGPT-Heavy Period | Impact |
|---|---|---|---|
| Task switches per hour | 8 | 13.5 | +68.75% |
| Average HRV (Lower indicates stress) | 72 ms | 57 ms | -20.83% |
| End-of-day fatigue rating (1-10 scale) | 4.1 | 6.5 | +58.54% |
| Continuous session length (mins) | 25 mins | 38 mins | +52% |
The Underlying Cause: Cognitive Multitasking
Leveraging insights from a recent study titled “Your Brain on ChatGPT: Accumulation of Cognitive Debt when Using an AI Assistant for Essay Writing Task” (MIT, Cornell University, June 2025), we further examined ChatGPT sessions for signs of linguistic fatigue.
- Users tended to use shorter prompts and more passive constructions as sessions progressed
- Prompts became more repetitive and syntactically simpler over time
- Users expressing uncertainty (“not sure”, “maybe”, “how do I start”) increased 41% in the final third of sessions
Our reinterpretation showed that frequent task-switching can result in up to 40% productivity losses due to “attention residue.”
When integrated with our data, it implies ChatGPT users lose approximately 2 hours per day of deep, focused work due to continuous context-switching.
ChatGPT use requires constant decision-making, verification, and re-prompting, subtly increasing cognitive load beyond standard multitasking.
Actionable Recommendations
For Employers:
- Create “AI-Free” blocks of work time for uninterrupted deep tasks.
- Develop training programs emphasizing strategic prompt usage, minimizing unnecessary cognitive load.
For Workers:
- Limit frequent switching between ChatGPT and other tasks. Batch prompts strategically.
- Incorporate regular short breaks to mentally reset, reducing cumulative stress.
Sources Compiled
- “Your Brain on ChatGPT: Accumulation of Cognitive Debt when Using an AI Assistant for Essay Writing Task”
Full research in PDF:
- “Working Distractions: the cost and ways to overcome” https://bioscope.ucdavis.edu/2020/01/17/working-distractions/
- “Meet Context Switching, the #1 Productivity Killer in the Workplace” https://www.techsmith.com/blog/context-switching/
- “AI fatigue is real: How to cut through the noise with real outcomes” https://www.replicant.com/blog/ai-fatigue-is-real-how-to-cut-through-the-noise-with-real-outcomes
This Is Where We Are Heading To
“AI Burnout” highlights a critical behavioral and productivity issue as AI becomes embedded in daily workflows. Businesses and individuals must rethink how they engage with AI tools to preserve mental energy and sustainably harness AI-driven productivity.
