The People and Places Behind the Machine

Part 26 of 30 · Series date: July 5, 2026

An AI response appears on a screen, making the interaction feel almost weightless. Somewhere, machines run, power flows, buildings need care, and people do the work. The convenience at the interface can make these conditions easy to overlook.

To judge the ethics, we must look beyond the person using the tool. A benefit received in one place may depend on burdens carried elsewhere. That observation does not prove the technology is unjust. It tells us which questions must be asked before calling it progress.

The material setting

The International Energy Agency's 2025 report, Energy and AI, studies power demand from data centers and AI. It also looks at ways AI might help energy systems. Its scenarios are projections, not guarantees about a particular community or facility. IEA, Energy and AI.

To judge a local proposal, we need local facts. Power sources, cooling arrangements, water use, land impacts, employment, and public duties can differ greatly. A world estimate cannot replace the site's permit papers or operating rules.

The town considered below is hypothetical. A real proposal needs its own review. We would need notices, applications, agreements, and independent studies. The industry's reputation cannot answer those questions.

Two legitimate views

Supporters of infrastructure may point to investment, services, jobs, or broader benefits from the technology it supports. These claims deserve review on their actual terms. A community should not dismiss potential opportunity just because the industry is unfamiliar.

Residents may ask about resources, noise, emissions, public costs, and what happens if expectations fail. These questions are not by nature opposition to technology. They are part of judging whether the proposed arrangement is fair and durable.

The ethical task is to distinguish documented commitments from hopes. What is guaranteed, what is forecast, and who bears the risk if forecasts are wrong? A benefit promised to a broad region may not compensate the people facing a concentrated burden. On the other hand, a concern should be supported with evidence suitable to the actual design.

The worker behind the service

Labor also belongs in the inquiry. Making equipment, repairing it, and supporting it remain part of the story. Calling the final product intelligent does not erase that work. Questions about working conditions and fair compensation remain relevant even when the user encounters only software.

The people who supply or review training material belong in this inquiry too. Were private records used with proper permission? Were contributors treated fairly? These are questions to investigate, not accusations about every provider. An attractive interface does not answer them.

Compare consequences at the right scale

Worldwide power use tells us little about the water needs of one particular site. A local concern about noise cannot establish the total social value of AI. Each claim belongs at a scale where relevant evidence can test it. Moving between scales too quickly can make both advocacy and opposition misleading.

Imagine a town considering a proposed facility. Ask about the actual design. How much power and other support will it need, and when? What will the public pay? What local jobs will remain? Which duties can be enforced? Forecasts should be distinguished from commitments. Construction jobs should be distinguished from continuing jobs. Broad green goals do not replace the actual design and rules of a local site.

The same close review belongs to objections. A dramatic statistic from a different climate or cooling design may not apply. A concern can be legitimate without every circulating number being accurate. Residents strengthen their position when they ask clear questions rather than accept the most alarming available comparison.

Efficiency does not automatically reduce total use

Suppose a system uses half the energy per task. Demand may still grow enough to raise total energy use. The per-task improvement is real, and so is the increase in total demand. These statements are compatible. Keep efficiency separate from total resource use. One does not tell us the other.

The opposite mistake is to count costs without asking what the activity replaces or enables. A computational service might support a valuable outcome or displace a more resource-intensive process. We need evidence that it really does this. Digital activity is neither weightless nor wasteful simply because it is digital. A serious comparison includes the relevant alternative and the full arrangement, as far as the evidence allows.

Uncertainty should be visible. If an estimate depends on demand growth, power sources, or future technical changes, readers should know that. Care with details in the displayed number cannot eliminate uncertainty in the assumptions. The IEA report offers possible futures and analysis. It cannot replace local facts or promise that a forecast will come true.

A neighbor may never use the product

The person receiving an AI-generated answer chooses to use a service. A resident living near its infrastructure may have made no such choice. A worker maintaining equipment may receive only a small share of the eventual benefit. These people still belong within the moral account. Consent at the interface does not authorize every burden along the chain.

This does not mean every affected person must possess a veto over every project. Communities often have to balance shared systems and competing needs. State the burdens openly. Give public reasons. Assign duties to people and groups able to meet them. A private benefit may place heavy costs on a small group. Payment for losses, steps to reduce harm, and enforceable conditions may then matter.

Calling someone a neighbor asks more of us when they are distant or unseen in the product. Stewardship cannot stop at the convenience of the person holding the phone. A claim that AI serves humanity must count these people too. Their work and surroundings make the service possible. Some may never get its most praised benefits.

Stewardship at community scale

Christian stewardship can be understood as attention to neighbors and shared conditions, not simply private restraint. A person may use a tool responsibly at home while still supporting better rules for its infrastructure. Being careful about our own use cannot replace holding public decisions to account.

Likewise, rejecting every use of a tool does not automatically address the institutions determining its impacts. Communities need clear facts and a real say. They also need duties that can be enforced. Those conditions make consent more than a presentation followed by a vote.

The question to ask is where the benefit travels and where the burden remains. Following both paths makes the moral picture more complete. A stronger account of progress includes the people whose work makes the tool possible.