The Insider/
What AI saves, and what it costs in community engagement

Communities are affected by major infrastructure projects in very real ways. It’s why project developers and regulators from the resources and energy sector engage and consult with those communities. It’s also why the emergence of artificial intelligence (AI) involvement in that process deserves further inspection. 

For developers, AI promises real savings in time, money and effort. It’s my view that those savings come at a cost. AI tools can capture and consolidate what people say, but much of what matters lies in how and why it was said. Picking up on those nuances relies on critical thinking and the kind of experience practitioners build over years in the field. When that judgement is handed off without oversight, the outcomes that engagement were meant to deliver are undermined. It raises the question of whether project developers are counting both the savings and the costs. 

What AI can save 

To start with, the savings deserve consideration. A project undertaking genuine community engagement can attract hundreds of submissions along with interview notes, meeting minutes, and survey responses. Reading, transcribing, categorising, and summarising takes considerable time which AI can do quicker and cheaper than a person or entire team. In the UK, researchers working with a local planning authority found that planners spent more than 60 hours logging and summarising 320 submissions which AI completed in minutes with no noticeable difference in quality. With some Australian agencies already disclosing that they may use AI to analyse submissions, it’s fair to say that using these tools is becoming commonplace; and that a first pass through large volumes of material is where AI offers the clearest value. 

Communities stand to gain too. Development applications and review documents can run hundreds of pages in length, so AI tools can help community groups work through dense and technical material, find the sections that affect or interest them most, and frame or write their responses accordingly. 

 The cost of savings 

The counterpoint is that the hours saved on reading are hours lost on listening. When a project developer hands over analysis to an AI tool, the output is a summary of what people said, not how it was said or why. The analysis quickly loses the understanding that comes from the actual reading. Going one step further, some platforms now offer AI-generated personas, profiles built to imitate a person or stand in for a community. Personas save the time and expense of finding real people, but the cost is the real people themselves and the conversations and storytelling that only in-person engagement can evoke.  

But it cuts both ways. There have been recent reports of planning submissions in the UK and submissions to parliamentary inquiries here in Australia that appear to contain AI-generated references. 

So, when a submission drafted by AI is summarised by AI and weighed against the views of an AI persona, it’s fair to ask who is talking to whom? 

Recent data breaches have made people more alert to whom holds their information. For First Nations peoples, the question carries more weight, and it’s one that researchers are already raising. An Indigenous-led study from Macquarie University recommended protecting cultural knowledge from extraction, misuse, and unauthorised inclusion in AI systems. The Australian Institute of Aboriginal and Torres Strait Islander Studies Code of Ethics recognises First Nations peoples’ right to control their culture and heritage, and to manage the collection, interpretation, and use of their information. Knowledge shared during engagement is shared on particular terms, and an AI tool doesn’t necessarily know what those terms are. Using one on knowledge shared on Country without explicit agreement risks taking control away from the people it belongs to. 

Counting both 

The challenge is that the savings and the costs don’t arrive at the same time and in the same way. The savings are counted in time and money which show up immediately whereas the costs are counted in trust and credibility which show up later. This matters because trust is built on engagement quality and procedural fairness, so a process that saves time by reducing human contact risks costing the thing it was meant to build.  

This isn’t an argument against the use of AI. It’s an argument for counting the cost of every nominal saving. AI earns its value where it handles volume and loses value where it replaces listening. As we looked at in a previous edition of The Insider, it’s like mistaking information for engagement. Perhaps reinvesting the time and money AI saves into meaningful conversations with communities is a good place to start. 

At ReGen Strategic, we specialise in community engagement, social impact assessments, and social performance. We start from the same premise that major development projects affect real communities in real ways. We see AI as a tool that augments good engagement practice, not replaces it. Our experience in designing and delivering community engagement strategies positions us to help clients judge where AI genuinely adds value and how community consultation should be handled when using AI tools. Our role is to make sure the savings realised in using AI tools aren’t at the cost of the relationships that engagement is meant to build. 

If you would like to walk through what that looks like for your community engagement, get in touch.