Context:
Developing a mindful approach to using AI in design that considers and reduces environmental and business costs
My Role:
Researcher, practitioner, and course developer
The costs of using AI
AI has become a default tool in design work. But defaulting comes with costs that rarely get discussed.
Environmental cost
Every prompt sends a request to a large model running on energy-hungry infrastructure that consumes significant amounts of water for cooling. Vague prompts, unnecessary follow-ups, and always-on background features compound quietly — across individuals, teams, and organisations — into a meaningful environmental footprint.
Business cost
Token usage, API calls, and AI tooling subscriptions add up fast — especially when prompts are poorly structured, outputs need heavy reworking, and AI is applied to tasks that don’t need it. Inefficient AI use isn’t just an environmental problem; it’s a cost problem.
The guidance gap
Most sustainable guidance asks designers to measure energy consumption, calculate carbon output, and report on usage. That’s a high bar that most practitioners will never clear — and it puts the burden in the wrong place.
My approach to efficient and mindful AI
My goal is a practice that is lighter on resources, lower in cost, and more considered in output—without adding significant overhead to how I already work.
Rather than building a new framework from scratch, this work maps existing mindsets onto AI use:
Sustainable thinking
Understanding the real-world environmental cost of AI, and designing use patterns that reduce it without eliminating the benefit.
Circular thinking
Borrowing from the refuse, reduce, reuse, remake, recycle hierarchy. Before using AI, ask if it’s needed. Before writing a new prompt, reuse or adapt an existing one. Before generating broadly, define precisely what you need and remake outputs rather than regenerating from scratch.
Inclusive thinking
Recognising that AI outputs reflect the biases of their training data, and building review practices that check for exclusion, stereotyping, and cultural blind spots before anything reaches a user.
How I developed this approach
I started by researching the environmental impact of AI—energy use, water consumption for data centre cooling, and the carbon cost of large model inference. Not to produce calculations, but to understand the scale well enough to make informed decisions about when and how to use it.
From there I researched existing practices around prompt efficiency and bias reduction, drawing on emerging guidance from the sustainable UX and responsible AI communities. I then looked at where circular design principles—already part of my broader practice—could be applied directly to AI use patterns.
The result is less a fixed methodology and more an evolving personal practice, one I’ve woven into my own workflow and developed into teachable content for the Sustainable UX course at LCD Academy.
To be clear about what this is: its a mindset shift, not a measurement system. It doesn’t have a controlled study behind it, and the evidence base for specific efficiency gains is still developing across the industry. What it does have is grounding in sustainable design thinking, circular principles, and a genuine understanding of where AI’s environmental and social costs sit.
1. Efficient AI in my own workflow
PROBLEM: AI tools are now part of everyday design work — research, writing, ideation, code. The temptation is to use them freely and frequently, treating generation as cheap. But every unnecessary prompt, every vague request that needs five follow-ups, every background feature quietly running in a design tool—these add up in energy, water, cost, and cognitive noise.
PROBLEM
Every vague prompt, unnecessary request, and idle background feature adds cost in energy, water, and money.
WHAT I DO: As a sustainable UX designer, my goal isn’t to reject AI, but to use it wisely, balancing creativity, care, and planetary responsibility. I aim to integrate AI responsibly into my workflow to reduce energy use, promote inclusivity, and keep my design decisions ethical.
Question if you need to use it
Choose lighter tools and formats
Prompt efficiently
Review for bias and accuracy
2. Designing AI-powered products
PROBLEM: AI is being built into almost every kind of digital product — recommendations, personalisation, chatbots, predictive interfaces. These features can make products smarter and more useful. They can also create overuse, waste, bias, and dependence if the design doesn’t account for their real cost — to users, to marginalised communities, and to the planet.
Most product teams focus on what AI can do. Fewer ask what it should do, how lightly it should run, and who it might exclude or harm.
PROBLEM
Pressure to embed AI into products is pushing teams to default to it, nudge users toward it, and ship it before asking if it’s needed, creating real costs in energy, bias, and user dependency
WHAT I DO: As a sustainable UX designer, I aim to guide how AI shows up in people’s lives, not only for convenience or profit, but for wellbeing, equity, and planetary care. This means questioning if it is even needed.
Question if it’s needed
Design it lightweight
Make AI visible and adjustable
Support agency, not dependency