Conservationists and scientists are sounding the alarm about a proposal from the mining lobby to employ artificial intelligence (AI) in expediting national environmental approvals. They warn that this approach could lead to serious failures similar to the “robodebt” scandal, putting already threatened species at greater risk.
The Minerals Council of Australia has urged the government to allocate $13 million for a trial that would use AI to assist companies in preparing application submissions and aid the federal government in decision-making.
According to the Biodiversity Council, a panel of independent experts from 11 universities, there are concerns that while AI could manage straightforward tasks, automating environmental assessments could lead to errors reminiscent of the robodebt scheme, where flawed computer-generated decisions are made without transparency. Such errors may push vulnerable species closer to extinction.
Sign up for the Breaking News Australia email
Robodebt pertains to the automated debt recovery program that wrongly accused hundreds of thousands of welfare recipients of overpayments between 2015 and 2019.
Lis Ashby, who leads policy and innovation at the Biodiversity Council, expressed concerns regarding Australia’s primary environmental law, the Environment Protection and Biodiversity Conservation Act. She stated that it is “filled with vague language and broad ministerial discretion.”
“These ambiguous rules prolong the assessment process as they hinder consistent, rules-based decision-making by human assessors. This lack of clarity will pose an even greater challenge for any AI tool,” she explained.
“Establishing clearly defined rules within the National Environmental Standards would expedite assessment timelines, even independent of AI, and is crucial for any future AI implementation,” she added.
Brendan Sydes, the national biodiversity policy advisor at the Australian Conservation Foundation, expressed skepticism about the Minerals Council’s initiative. “While technology can enhance the efficiency of nature protection laws, it is important to remember that AI can be a poor master,” he remarked.
He emphasized that the federal government should instead focus on addressing existing gaps in data concerning threatened species and their habitats.
Professor David Lindenmayer, a forest ecologist from the Australian National University and a member of the Biodiversity Council, noted that one-third of Australia’s threatened species have not been adequately monitored, while others only have incomplete data.
Assessors often compensate for these data deficiencies through expert consultations.
“AI-driven decisions depend on the quality of available data, and for most of Australia’s threatened species, reliable information — like basic location data — is not publicly accessible,” he said. “Automating decisions with flawed or outdated data threatens to undermine biodiversity protection.”
Last year, the Albanese government implemented reforms to environmental legislation after a 2020 review revealed that existing measures failed to adequately protect species and ecosystems.
Professor Hugh Possingham, a noted conservation biologist at the University of Queensland, stated, “AI tools typically require training materials to function effectively. However, the past 20 years of EPBC Act approvals are unsuitable due to their evident failures in environmental protection.”
He advocated for the employment of additional personnel to conduct assessments to expedite the process instead.
Tania Constable, chief executive of the Minerals Council, expressed disappointment over the negative comparisons to robodebt, asserting that the proposal is both innovative and could enhance environmental protection while improving overall efficiency.
She explained, “Our approach aims to augment human decision-making with AI tools to assist both regulators and project proponents in navigating the complexities of assessments under the EPBC Act.”
A spokesperson from the federal government indicated that decisions regarding budget allocations would be made in due course, though the environment department is currently assessing how AI could facilitate easier application processes.
“Ultimately, decisions on project approvals will remain the responsibility of assessment officers, not AI,” the spokesperson clarified. They also noted that AI tools could potentially streamline processes, reduce uncertainties, and translate technical jargon.
