Report urges use of sovereign AI for environmental approvalsA new report makes a powerful argument that the judicious use of AI can make it possible both to maintain high environmental standards and to make the approval process faster and easier.Tim JohnstonThe use of sovereign AI can speed up environmental approvalsGift this article4 min read2 hours agoA new report has called on the government to use more Australian sovereign artificial intelligence to streamline Australia’s expensive and slow environmental approvals process.The report draws on international examples to suggest tens of millions of dollars in potential savings for industry and governments, and dramatic time savings for the $278bn in construction projects presently in the approvals pipeline. The report, from the Committee for Economic Development of Australia in partnership with Brisbane-headquartered software developer TechnologyOne, says the current system of manual assessment is breaking under the weight of more and more complex environmental applications, just as the country feels the need to accelerate investment.“Revisiting the tools, technology and systems we use to understand the impacts of development offers the potential to support productivity growth while building a deeper understanding of how economic activity impacts our environment,” the report says.“Getting this right means faster delivery of affordable housing, generating clean energy and building the transmission to deliver it, tapping into the economic opportunity of critical minerals and enhancing our ability to sustainably build the data centres that can give us an edge in the age of AI.”In the political lexicon, “streamlining” environmental legislation is often seen as shorthand for making it easier to pollute, degrade and destroy our environment. However, the CEDA report makes a powerful argument that the judicious use of AI can make it possible both to maintain high environmental standards and to make it faster and easier for project developers to get approval.Australia’s environmental approvals problem is substantial, metaphorically and literally. The Environmental Impact Statement for Australia Pacific LNG’s gas project in Queensland took a year to prepare and runs to just over 10,000 pages. If anyone was environmentally unsound enough to print it out, it would stand a metre high and weigh as much as 50kg or half a dozen koalas. Even 10 years ago – long before AI – the Red Tape Committee concluded that streamlining federal and state environmental approval processes could save Australian businesses $426m a year. Other countries are already moving towards greater AI assistance in the planning process. Denmark’s AI-assisted environmental assessment platform is projected to save $80m annually, and three English councils cut submission processing time by 66 per cent using natural language processing. The government recognises that something needs to be done. In his latest budget, Treasurer Jim Chalmers announced $105.9m to improve the experience for those applying for environmental approval, including through the use of AI, and better access to environmental data. At heart, Australia is a resource-based economy with absurd housing prices that needs a faster and more efficient development process, but it is also a uniquely delicate environment that supports communities whose interests have been too-often sidelined. The resulting tension creates a deeply complex interplay of priorities, says Liam Dillon, the author of the CEDA report. “Complexity is a feature of the system, not a bug, because it’s important that we understand the full range of environmental impacts,” he says, pointing to the voluminous legislation and multiple agencies involved. CEDA says that it costs more than $250,000 in time alone for the government agencies to review a major environmental impact submission manually: the government’s response to the submission from APLNG ran to 345 pages. “Assessment officers must work through voluminous documentation, cross-reference it against regulatory requirements, identify information gaps and co-ordinate with other agencies to prepare recommendations that are legally defensible, transparent and consistent,” the report says.Dillon says AI’s ability to manage large data sets would make a significant difference.“So this is where we see the opportunity for AI to come in: what it’s really good at is sifting through complexity and identifying key trends, and when done well, that can expedite the process without sacrificing the integrity of analysis,” says Dillon, CEDA’s head of economics. He warns that although AI can streamline the planning approvals process, it is unlikely to be able to carry out some of its most crucial functions for the foreseeable future. He envisions a series of tools that carry out specific subsets of the approvals process.“Identifying where these tools are appropriate, but more importantly, where they’re not, is vital for developers and governments alike,” says Dillon. “What you really don’t want to be doing is finding yourself in a situation where you’re just accelerating existing inefficiencies.” He says AI would be well-suited to see if a project’s technical drawings are compliant with planning regulations for example, but humans would still have to be central to evaluating less-clearly defined parts of the process like community engagement or cultural issues. And, he says, that points to an important caveat. “We know that within the environmental approvals space, highly localised environmental data is very important because when you’re using models that are trained on maybe data from Europe or the US, for example, it’s just an extra set of complexity that doesn’t really reflect sort of local conditions” and can amplify AI hallucinations. He says the government should use at least some of its new budget to outsource the development to Australian commercial partners. State and federal governments are struggling to attract IT workers, but Dillon suggests established Australian technology providers, particularly those with a track record working with regulators and councils, are well-placed to help fill the gap. “Empirically, what you see is that where you’re able to use more local and domain-specific knowledge, the models just produce better results. The second thing relates to data sovereignty,” he says, pointing out that many government departments are limited to using Australian products for good reason. “There are risks in terms of the sensitivity of the data that they handle. You could think of really large infrastructure projects, potentially in the energy space or port infrastructure, for example, that contain both commercially sensitive but also strategically sensitive information.”Australia has both abundant resources and astonishing environmental riches. AI offers an opportunity to create an environmental approvals process that is both fast and effective, boosting the economy while protecting our green heritage.This content was produced in association with TechnologyOne. Read our policy on commercial content here.Read related topics:Artificial Intelligence