The Firewalkers guide to responsible, sustainable and wise technology — and how to use AI as a tool for civic action and community protection.
The Firewalkers use AI. We celebrate innovation. We believe technology has a vital role in solving the greatest challenges of our time. The question is not whether AI is good. The question is: can we build an AI future that enriches life rather than extracting from it? This page is our answer — and a practical toolkit for anyone who shares that question.
These principles apply to AI systems, AI infrastructure, and the organisations that build them. They are not aspirational. They are the standard we hold, and the questions we bring to every development decision.
Good AI is evaluated not only on profitability but on what it does for human flourishing, ecological health and long-term community wellbeing. Economic value matters. Ecological value matters. Both belong in the assessment.
Good AI infrastructure accounts transparently for energy consumption, water use, land use, rare earth extraction, and carbon emissions — across the full lifecycle, not just at the point of delivery.
Good AI runs on renewable energy, pursues efficiency as a design goal rather than an afterthought, and commits to decarbonisation timelines that are real and verifiable — not greenwash.
Good AI is built on open standards where possible, supports the right to repair, does not extract from public data while excluding public benefit, and treats the digital commons as a shared resource rather than a private resource to be enclosed.
Good AI infrastructure is sited with genuine community involvement, not just procedural consultation. Local knowledge, local impact, local voice. Communities affected by AI infrastructure development are stakeholders, not obstacles.
Good AI infrastructure is built to meet verified, real-world need — not speculative forecasts driven by capital allocation cycles. The lesson of the railway mania is not to stop building. It is to build to actual demand, not projected bubbles.
Good AI development creates genuine value for the communities and landscapes that host its infrastructure — through employment, energy sharing, community benefit funds, ecological restoration, or other meaningful returns.
Good AI thinking extends beyond the next quarter. Future generations are stakeholders. The decisions we make about AI infrastructure today will shape landscapes, energy systems and communities for decades. That weight belongs in every boardroom and planning committee.
"Is this the right infrastructure, in the right place, at the right scale, for the right future?"
The Firewalkers — the question we bring to every developmentThis single question is the most useful tool in the Firewalkers toolkit. It does not oppose technology. It does not oppose data centres. It demands that every proposal justify itself against the future it is supposed to serve — not the future of the investors, but the future of the place, the community, and the people who will live there.
The Firewalkers do not oppose data centres. We make distinctions. The same distinction a land steward makes between a well-tended field and a strip-mined hillside. Both are agriculture. The difference is whether you plan to come back.
Copy any of these prompts and paste them into Claude, ChatGPT, Gemini or any AI assistant. They are designed to help you understand, campaign and act effectively. Replace the bracketed text with your specific details.
Decode planning documents in minutes — including what they are designed to obscure.
A well-argued letter to a planning authority or MP carries real weight when it is specific, evidenced and well-structured.
Environmental Impact Assessments are designed to be read by regulators, not residents. AI can translate them.
Most people do not know how much power they actually have in planning and regulatory processes.
The strongest campaigns propose better solutions, not just objections. AI can research international best practice fast.
Campaigns win on evidence. AI can help you quickly assemble the data, statistics and precedents you need.
Turn complex technical proposals into clear, accessible summaries that your whole community can understand and act on.
Understand what a development means for local biodiversity, water catchments and ecological connectivity.
The largest AI models require enormous energy to run. But smaller, more efficient models have improved dramatically — many tasks that seemed to need the biggest models can now be done with models 10 to 100 times smaller, at a fraction of the energy cost. And some can run entirely on your own hardware.
European-built, fully open source. Mistral 7B and Mistral Nemo perform comparably to models many times larger. One of the most energy-efficient capable models available. Good first choice for civic campaigns.
mistral.ai →Specifically designed to be small but highly capable. Phi-3 and Phi-4 can run on a laptop. Exceptional performance per watt. Good for document analysis, drafting and research without cloud dependency.
Microsoft Phi →Available in sizes from 8B to 70B parameters. Fully open source — can be downloaded, run locally and customised. The 8B version runs well on consumer hardware and is sufficient for most campaign tasks.
llama.meta.com →Running AI locally means no data is sent to a third party, no cloud costs, and the model runs whether or not you have internet access. For sensitive campaign work — or simply for communities that want control over their own tools — local AI is increasingly practical.
The simplest way to run open-source models on your own computer. One command to install. Works on Mac, Windows, Linux. Run Mistral, LLaMA, Phi and dozens of others entirely locally. Free. Private.
ollama.ai →Build a private AI assistant on your own files — PDFs, planning documents, ecological reports, meeting notes. Ask questions of your own document library. Runs locally. Your documents never leave your machine.
anythingllm.com →A user-friendly desktop interface for running local models. Download models, chat with them, use them to analyse documents — all with a clean interface, no command line required. Good starting point for non-technical users.
lmstudio.ai →The Firewalkers teach people to use AI as a tool for justice and protection. Here is how to use it well.
AI can make mistakes, misread documents, or generate plausible-sounding but incorrect statistics. Always check key facts against primary sources before using them in a formal submission.
The more specific your prompt, the more useful the output. Include the location, the type of development, the specific concern, the audience. Vague prompts produce vague answers.
AI-generated letters are a starting point, not a finished document. Add your personal experience, your local knowledge, your voice. Personal testimony carries weight that generic text does not.
When AI gives you a statistic or a claim, ask: "What is your source for this?" Then verify it. Good AI use means treating the output as a research assistant, not an oracle.
If the first answer is not quite right, say so. "That's good but can you make it more formal / add a section on water use / cut it to one page." AI responds well to specific feedback.
If you develop a prompt that works particularly well for your local context, share it at the Firewalkers hearth. The movement is built from accumulated practical knowledge.
Join the Firewalkers community — new resources, campaign updates, and a growing network of people using technology wisely to protect the places they love.
Keep me informed