A new report calls on the government to make greater use of Australia’s sovereign artificial intelligence to streamline Australia’s expensive and time-consuming environmental approvals process.
The report cites international examples and suggests potential savings for industry and governments in the tens of millions of dollars and dramatic time savings of $278 billion for construction projects currently in the approval pipeline.
The report, released by the Australian Economic Development Board in partnership with Brisbane-based software developer TechnologyOne, says current manual rating systems are buckling under the weight of increasingly complex environmental applications, just as Australia feels the need to accelerate investment.
“Rethinking the tools, technologies, and systems we use to understand the impacts of development offers the potential to support increased productivity while developing a deeper understanding of how economic activity impacts the environment,” the report says.
“Getting this right means strengthening our ability to deliver affordable housing faster, build the grid to generate and deliver clean energy, capitalize on the economic opportunities of critical minerals, and sustainably build data centers that give us an advantage in the age of AI.”
In political parlance, “rationalization” of environmental law is often seen as shorthand for making it easier to pollute, degrade, and destroy the environment. However, the CEDA report makes a strong case that using AI wisely can help maintain high environmental standards and help project developers get approvals faster and easier..
Australia’s environmental approval issues are significant, both figuratively and literally.
The environmental impact report for Australia Pacific LNG’s gas project in Queensland took a year to prepare and is just over 10,000 pages long.
If someone who wasn’t environmentally conscious printed it, it would be 1 meter tall and weigh 50 kg, or the weight of six koalas. Even a decade ago, long before AI, the Red Tape Commission concluded that Australian businesses could save $426 million a year by streamlining federal and state environmental approval processes.
Other countries are already moving to expand AI support in the planning process. Denmark’s AI-assisted environmental assessment platform is predicted to save $80 million a year, and three UK councils reduced submission processing times by 66% using natural language processing.
The government recognizes that something needs to be done. Finance Minister Jim Chalmers announced in his latest budget that $105.9 million will be spent to improve the experience for environmental approval applicants, including the use of AI and improved access to environmental data.
Australia is essentially a resource-based economy, with exorbitant house prices and a need for faster and more efficient development processes, but it is also a uniquely sensitive environment in which to support communities whose interests have been too often sidelined.
The resulting tensions create a very complex interplay of priorities, says Liam Dillon, author of the CEDA report.
“Complexity is a feature of the system, not a bug, because it’s important to understand all the environmental impacts,” he said, noting the vast amount of legislation and multiple agencies involved.
CEDA says it costs the agency more than $250,000 in time alone to manually review key environmental submissions. The government’s response to APLNG’s submissions is 345 pages long.
“Assessors must sift through extensive documentation, cross-reference it against regulatory requirements, identify gaps in information, and collaborate with other agencies to develop recommendations that are legally defensible, transparent, and consistent,” the report states.
Dillon says AI’s ability to manage large data sets will make a big difference.
“That’s where we see an opportunity for AI to come in. What AI is really good at is sifting through complexity and identifying key trends, and when done well, it can speed up the process without sacrificing the integrity of the analysis,” said Dillon, CEDA’s head of economics.
He cautions that while AI can streamline the plan approval process, some of its most important functions are unlikely to be performed for the foreseeable future. He envisions a set of tools to perform a specific subset of the approval process.
“Identifying where these tools are appropriate, and more importantly, where they are not, is important for developers and governments alike,” says Dillon.
“What you really don’t want to do is end up in a situation where you’re only accelerating existing inefficiencies.”
AI is good at, for example, checking whether a project’s technical drawings comply with planning regulations, but humans still need to be at the center of assessing less clearly defined parts of the process, such as community engagement and cultural issues, he says. And that points to an important caveat, he says.
“We know that in the field of environmental approvals, highly localized environmental data is very important, because if you have a model trained on data from, say, Europe or the United States, it’s just an extra set of complexities that don’t really reflect the local situation,” and can amplify the illusions of AI.
He argues the government should use at least part of the new budget to outsource development to Australian commercial partners. While state and federal governments are struggling to find IT talent, Mr Dillon suggests Australia’s established technology providers, particularly those with a track record of working with regulators and parliament, are well placed to fill the gap.
“What we see from experience is that models only produce better results when you can use more local, domain-specific knowledge. The second thing has to do with data sovereignty,” he said, noting that many government departments are restricted to using Australian products for good reason.
“There are risks in terms of the sensitivity of the data that we deal with. For example, you can think of very large infrastructure projects, such as in the energy sector or port infrastructure, where there is not only commercially sensitive information, but also strategically sensitive information.”
Australia has abundant resources and an incredibly rich environment. AI offers the opportunity to create a fast and effective environmental approval process, promoting the economy while protecting our environmental heritage.
This content was created in partnership with TechnologyOne. Read our commercial content policy here.
