Responsible Use
Cost to the Environment
Original work: "Educators' guide to multimodal learning and Generative AI" — Tünde Varga-Atkins, Samuel Saunders, et al. (2024/25) — CC BY-NC 4.0
Adapted for UK Nursing Education by: Lincoln Gombedza, RN (LD)
The environmental cost of GenAI is often invisible but significant. Each individual prompt uses little energy; the cost comes from scale, from training ever-larger models, and from image and video generation. As nursing professionals committed to health and wellbeing, understanding and mitigating these environmental impacts is part of our professional responsibility.
The Hidden Energy Costs
Training Models
Massive Consumption: Training GPT-3 (2020) was estimated at about 1,287 MWh of electricity (Patterson et al., 2021), roughly the annual use of 120 US homes. Today's frontier models are far larger, and providers rarely publish training figures.
Water Usage: Data centres use water for cooling. Training GPT-3 in Microsoft's US data centres was estimated to evaporate about 700,000 litres of fresh water on site, and 5.4 million litres including electricity generation (Li et al., 2025).
Daily Usage
Query Cost: Providers now publish per-prompt estimates: about 0.24 Wh and 0.26 ml of water for a median Gemini text prompt (Google, August 2025) and about 0.34 Wh for an average ChatGPT query (OpenAI, June 2025). These are company figures, not independently audited, and longer reasoning or agentic tasks use much more.
The bigger picture: Data centres used about 415 TWh of electricity in 2024 (around 1.5% of global demand), projected to more than double to about 945 TWh by 2030, driven mainly by AI (IEA, 2025).
Media Impact: Generating images uses far more energy than generating text, and video more again (Luccioni et al., 2024).
Carbon & E-Waste
💨 Carbon Footprint
Direct: 24/7 Data centre operations and cooling.
Indirect: Manufacturing specialized GPUs and transporting equipment globally.
🗑️ E-Waste (Hardware)
Rapid Obsolescence: AI drives demand for powerful new devices, shortening lifecycles.
Disposal: E-waste contains toxic materials and is difficult to recycle responsibly.
Nursing's Responsibility
Future Nurse proficiencies, Platform 2: Promoting health and preventing ill health.
Environmental health is public health. Nurses have a duty to minimise harm, and sustainable practice is part of holistic care.
NHS context: The NHS has committed to net zero by 2040 for the emissions it controls directly, and by 2045 for the emissions it influences, including its supply chain (Greener NHS). Digital and AI services the NHS buys fall within that supply chain.
Mitigation Strategies
For Educators
Conscious AI Use
- Evaluate Necessity: Ask "Is AI the best tool?" before using.
- Efficient Tools: Use smaller models or text-only for simple tasks.
- Optimise: Craft effective prompts to avoid retries. Archive and reuse responses.
Institutional Actions
- Ask suppliers for energy, carbon and water data, and prefer providers that publish it.
- Include AI use in sustainability reporting where it is significant.
For Students
Mindful Consumption
- Reduce: Don't use AI for tasks you can easily do yourself.
- Optimize: Plan prompts to be specific and reduce iterations.
- Stewardship: Maintain devices to extend their lifespan.
Practical Examples
❌ High-Impact (Avoid)
- Using AI for simple factual lookups.
- Generating unnecessary videos/images.
- Replacing hands-on practice with AI sims.
- Constantly upgrading hardware for AI.
✅ Sustainable (Adopt)
- Hybrid: Draft with specific prompts, refine manually.
- Low-Tech: Peer discussion & library resources.
- Share: Distribute AI summaries to the cohort to save everyone regenerating.
Reflection Questions
🌍 Sustainability Check
- Awareness: Which of your AI tasks are energy-heavy (image, video, long reasoning), and which are light (short text)?
- Necessity: Which interactions could be replaced with lower-impact alternatives?
- Advocacy: How can you promote green AI in your institution?
Next: Understand Cost to Knowledge and how AI affects learning and retention.