In Canada and other countries, governments are trying to build or develop more housing, transmission lines, mines, energy projects and data centres while simultaneously expanding public participation in the decisions that shape them.

Artificial intelligence may be about to make that combination much harder to manage.

Toronto, Mississauga and Hamilton, as well as other municipalities across Canada, have recently wrestled with data centre development.

Somewhere in one of those debates, a resident has almost certainly uploaded a planning report into ChatGPT or a similar AI tool and typed a simple prompt:

“Help me write a five-minute presentation opposing this project.”

Seconds later, they had a polished argument, technical questions and a set of objections tailored to the data centre proposal.

The irony of using AI to oppose the infrastructure that powers AI is obvious. But the more important story is what this technology may do to the approval process itself.

Governments have spent years trying to make public consultation more accessible. AI may become one of the most powerful tools ever created for doing exactly that.

But it creates an asymmetry that policymakers have barely begun to consider.

AI lowers the cost of participation

The time and cost of creating sophisticated public submissions are collapsing. The time and cost of evaluating them by municipalities and their staff are not.

Governments should therefore modernize the other side of the equation, creating greater balance, by developing AI tools that help approval authorities sort and assess this growing AI-generated input.

Canada is trying to build more major infrastructure projects at precisely the moment AI is making it dramatically easier for citizens to participate in, and oppose, the approval of those projects.

For decades, public consultation contained a natural constraint. Preparing a serious intervention took time. Residents had to read hundreds of pages of technical material, understand planning rules, assemble evidence and organize their arguments. Some hired lawyers, planners or environmental consultants.

Those barriers were hardly democratic virtues. They meant that people or companies with more time, money and expertise often had a louder voice.

Generative AI changes that dramatically.

Every citizen now has access to something approaching a researcher, policy analyst and communications adviser. AI can summarize in minutes an environmental study, identify weaknesses in a planning report, compare a proposal with municipal policy, draft a presentation and prepare questions for elected officials.

The cost of producing sophisticated public participation is approaching zero. That is in many respects a good thing.

One campaign, 1,000 unique submissions

However, while a resident can now generate a detailed 10-page submission in minutes, a municipal planner still has a professional obligation to read it, determine whether its claims are accurate, distinguish legitimate planning arguments from irrelevant ones and incorporate appropriate issues into a staff report.

Multiply that by thousands of participants.

AI also changes something important about mass participation.

Social media already makes it easy for campaigns to distribute form letters and encourage hundreds or thousands of people to submit them. Authorities can usually recognize these for what they are – many people expressing support for the same argument. They can count the participants while treating the substantive issue as one recurring concern.

However, the same campaign can now provide supporters with a prompt rather than a form letter. Each person can generate a unique submission – different language, different examples, different citations and slightly different arguments – even when all of them originate from the same underlying campaign.

Suddenly, what would once have appeared as 1,000 copies of the same letter can arrive as 1,000 apparently independent submissions.

That raises difficult administrative questions. Does each require separate analysis? How should governments distinguish between the number of people expressing a concern and the number of genuinely distinct issues being raised? How do we preserve meaningful public participation without allowing sheer information volume to overwhelm the institutions responsible for making decisions?

Canadian governments already acknowledge that approvals for megaprojects take too long. Ottawa and the provinces are creating major project offices, promising faster approvals and attempting to reduce duplication. On Sept. 21, the federal government introduced Bill C-39, the Building Canada Strong Act which proposes completing federal project reviews within a year.

At the same time, the amount of material entering those approval systems may be about to increase dramatically.

The challenge is not simply volume. AI can also make submissions look more authoritative than they are. Its argument may contain genuine insight or it may contain misunderstood evidence, invented precedent and conclusions that sound technical but are wrong.

The burden of determining which is which still falls on human officials.

This is no longer entirely theoretical. In Britain, AI-powered services are already being used to analyze planning applications and generate objections, while planning authorities are experimenting with AI tools of their own to process increasingly large volumes of public input. The beginnings of an AI arms race between applicants, opponents and regulators are already visible.

Canada’s AI strategy needs a smarter build-buy approach
New federal agencies won’t fix old problems with delivery
Canada needs to develop its own AI computing power

In Canada, data centres may simply be the first place we see this dynamic clearly because AI provides such an irresistible organizing process. The technology itself is generating demand for infrastructure while simultaneously giving opponents of that infrastructure increasingly powerful tools.

The same pattern applies to other megaprojects.

A resident opposing a transmission line can ask AI to analyze electromagnetic field studies and route alternatives. Someone fighting a quarry can have AI review hydrogeological reports. A neighbourhood group opposing a housing development can ask AI to identify every potentially applicable provision in an official plan.

AI is not creating the underlying objections. It is removing the previous difficulties in researching, packaging and communicating them.

Governments need AI tools of their own

The answer cannot be to restrict public participation. Governments should instead modernize the other side of the equation.

At a minimum, governments should begin developing AI tools that can help authorities:

• sort and categorize large volumes of submissions
• separate recurring arguments from genuinely new issues
• verify citations and flag questionable factual claims
• connect comments to relevant planning or regulatory criteria
• distinguish mass participation from genuinely distinct substantive arguments
• summarize consultation without losing minority viewpoints
• provide transparent records showing how significant concerns were addressed

Human decision-makers must remain accountable for the conclusions. But there is little reason they should continue manually processing information that AI can organize far more efficiently.

Governments should also reconsider how consultation itself is structured. The rules governing participation were designed for a world in which generating a detailed submission required meaningful human effort. That assumption is disappearing.

The challenge will be to preserve what matters – the right of citizens to be heard, the identification of legitimate concerns and accountability for decisions – without requiring officials to treat every AI-generated variation of the same argument as an entirely new body of evidence.

AI is going to alter the relationship between citizens and the state. It will dramatically increase ordinary people’s ability to interact with complex government processes, challenge expert evidence and organize collective action.

Our institutions will have to become equally capable.

Canadian governments say they want to build faster. They also want citizens to participate meaningfully in the decisions that affect their communities.

Those goals are compatible, but only if the tools between public participation and government decision-making become dramatically more efficient.