People can hold two opinions about AI at the same time. They can see what it helps them do at work and at home and still object to the data center being proposed near them. In my October 19 conversation with Ken Lanphear on WKZO, we talked about how to give those people a voice in the decision and make the terms of a build mutually beneficial.
A person concerned about their water or electricity bill has a question about what the project will do to them. Explaining how useful AI is does not answer that question. We have to work through the proposal, including whether the people being asked to accommodate it will see the benefits and what they can do if the promises are missed.
That starts with knowing what is fixed and what can change. If the location has already been settled, people need to know that before a meeting advertised as a chance to shape the project. If a different cooling design, a smaller build or a later expansion is possible, those choices need to be available while people can still do something with them. Promising a negotiation over something that cannot change gives people another reason to distrust the next invitation.
We need a non-binary discussion long enough to work through those terms. A community also needs to be able to refuse a proposal when the terms cannot be made acceptable.
What people are being asked to trust
Our trust work gives us a way to understand why people can hear the same offer and have different concerns. Someone whose well or sleep could be affected needs an answer about that continuing exposure. Someone hoping to work on the construction needs to know whether the project will proceed and how long the work will last. A tax-revenue estimate does not settle either question on its own.
The trust circumplex we have discussed before helps us recognize legitimate priorities that pull against each other. Control over an outcome and freedom to change it can both be reasonable things to ask for. If we dismiss the protection another person needs, we can lose the chance to find terms that work for both of us.
Ken asked who would oversee a deal and who would bring people together. Our trust framework puts responsibility, independent verification, meaningful choice and correction into that discussion. A resident needs to know who owes the promise and who will investigate when it appears to have been missed. They also need a way to use the findings, whether that means changing an operation, obtaining relief or stopping a further phase.
In a low-trust political environment, the transparency needs to continue over time, after the vote. People will have more to judge if the promised reports arrive, the costs are visible and a complaint produces an answer.
What is each person being asked to rely on?
What previous builds have brought
Virginia’s legislative auditor, JLARC, found in its 2024 study that data centers brought substantial construction activity and, in some localities, significant tax revenue. Operating employment was much smaller than construction employment. The study also documented poor siting decisions near homes and problems with continuous, low-frequency noise. Most facilities did not generate noise complaints; location and design helped explain the difference.
A county can gain revenue while a neighborhood bears a cost that should have been addressed before construction. The construction work can be worthwhile even though most of those jobs end. Michigan needs its own estimate of the revenue left after incentives and public-service costs; Virginia’s tax rules and concentration of data centers cannot supply that answer for us.
A sound baseline, the location of equipment and the cooling design can be settled while alternatives are available. After the facility is operating, the resident may be trying to get relief from something that would have been easier to prevent.
Who can decide the terms?
Michigan’s Public Service Commission explains that its role in the DTE–Google special-contract case concerns utility service and its terms. It does not approve the data center’s location or decide water use.
The township meeting needs a clear account of who can decide the questions left to it and who handles the others. Residents need to know when those decisions will be made and how to participate.
Some commercial information will be confidential. Independent advisers can help examine it, but their access and their responsibility to report publicly need to be clear. Qualified participants seeing an unredacted utility contract under a protective order does not mean every resident can see it. The public still needs enough information to understand the commitment being made on its behalf.
The terms should cover an independently chosen technical review before the final decision, with enough time for residents to use it. If the applicant funds the work, it cannot control the findings.
Put the question in the room that can decide it
Where negotiation can go wrong
Game theory can help us examine what each side has to gain, what it knows and what it can credibly promise. The Virginia neighborhood problems raise a question about who gets counted. A bargain can look mutually beneficial to a developer and a locality collecting revenue while leaving the neighbor’s continuing noise burden out of the calculation. JLARC documented the planning problems; the bargaining interpretation is mine.
There is also a change in what each side can afford to do. Before approval, the community can compare other uses for the site and the developer can compare other locations. Once a utility has built infrastructure for the project, cancellation can leave a large cost behind. The terms need to assign that cost before the investment makes it harder to walk away.
A repeated relationship can encourage cooperation when people expect to need each other again. A sale, an election or a project company with little money can change that expectation. The agreement has to keep working when the people who made the promises are gone.
A bad outcome alone does not show that game theory failed. It can prompt us to ask whether a negotiation left people out, relied on information they could not check or assumed a promise would remain enforceable.
Staying at the table safely
People need enough information and independent help to challenge a proposal while it can still change. They need to be able to refuse it without having their concerns dismissed as opposition to the future. A community should also explain what evidence or changed terms could make the project acceptable, if any would, so an applicant can decide whether to redesign, relocate or stop pursuing that site.
Sitting out has consequences, including work and revenue that may go elsewhere. Accepting a project has consequences that people nearby may live with for decades. A national need for computing cannot settle the choice for a particular Michigan township, and using AI should not commit someone to accepting every facility proposed near them.
Before the next vote, ask which terms can still change, which body makes each decision and who can check whether a commitment was kept. The people who will live with the agreement need to be able to use those answers after the construction is over.