Me & AI
Discover the environmental and cognitive impact of your AI usage.
What research says
Cognitive Impact
Cognitive offloading weakens memory formation
When we routinely delegate thinking tasks to AI, our brains engage less in the encoding processes that form long-term memories. Research shows that 'cognitive offloading' — letting technology remember for us — reduces our own retention capacity.
Risko, E.F., & Gilbert, S.J. (2016). Cognitive Offloading. Trends in Cognitive Sciences, 20(9), 676-688.Automation complacency: trusting AI without verification
Decades of human factors research shows that humans develop 'automation complacency' — a tendency to over-trust automated systems and stop checking their outputs. With AI, this manifests as accepting AI-generated answers without critical evaluation.
Parasuraman, R., & Manzey, D.H. (2010). Complacency and Bias in Human Use of Automation. Human Factors, 52(3), 381-410.Dependency
Screen time over 4 hours daily linked to increased anxiety
Large-scale studies consistently find that screen-based digital activity exceeding 4 hours per day is associated with higher levels of anxiety and depressive symptoms, particularly in younger populations.
Twenge, J.M., et al. (2018). Clinical Psychological Science, 6(1), 3-17.AI tools may accelerate cognitive deskilling
When AI handles tasks we once performed manually — writing, calculating, navigating — the neural pathways for those skills can weaken through disuse, a phenomenon called 'cognitive deskilling'. This doesn't mean AI is harmful, but that intentional practice of core skills remains important.
UNESCO (2023). Guidance for generative AI in education and research.Energy & Environment
One ChatGPT query uses 10x more energy than a Google search
A single GPT-4 query consumes approximately 0.005 kWh — roughly ten times more electricity than a standard Google search. As AI usage scales globally, this difference becomes significant.
IEA (2024). Electricity 2024. International Energy Agency.AI data centers could consume as much electricity as Japan by 2026
The International Energy Agency projects that global data center electricity demand could more than double by 2026, largely driven by AI workloads, approaching the total consumption of Japan.
IEA (2024). Electricity 2024.Training GPT-4 consumed as much water as 700 average households use monthly
AI model training and inference require significant water for data center cooling. Microsoft's water consumption rose 34% in 2022, largely attributed to AI development. Each ChatGPT conversation of 20-50 questions uses approximately 500ml of water.
Li, P., et al. (2023). Making AI Less Thirsty. arXiv:2304.03271.Productivity
The productivity paradox: more output, less deep thinking
While AI undeniably speeds up content production, research suggests it may simultaneously reduce the quality of deep, focused thinking. Workers report producing more but understanding less of what they produce.
Carr, N. (2020). The Shallows: What the Internet Is Doing to Our Brains. Updated ed.Sleep & Health
Evening AI chat sessions can disrupt sleep patterns
Engaging in stimulating AI conversations before bedtime can delay sleep onset, similar to other screen-based activities. The cognitive engagement from AI interaction is often higher than passive scrolling, making it harder to wind down.
Based on screen-time and cognitive arousal research.Social Impact
AI-mediated communication changes how we relate to others
When people use AI to draft messages, emails, or social media posts, the communication becomes more polished but less authentic. Recipients may sense the difference, potentially eroding trust in digital communication.
APA (2023). Health Advisory on Social Media Use in Adolescence.Want automatic tracking? Install our Chrome extension.
Download extensionWant to learn more about your relationship with AI?
Take the self-assessment quiz