Using AI to speed up Australia’s environmental approvals risks ‘robodebt-style’ failures, scientists say | Australia news

Environmentalists and scientists have warned that a proposal by the mining lobby to use artificial intelligence to speed up national environmental approvals could generate “robot-debt” failures, putting endangered species at risk.

The Minerals Council of Australia has called on the government to spend $13 million to test the use of AI to help companies prepare applications and help the federal government make decisions.

But the Biodiversity Council, a group of independent experts from 11 universities, told Guardian Australia that while AI could play a role in simple tasks, automating environmental assessments “could lead to robodebt-style failure, where computers make wrong decisions without transparency” that could ultimately push species closer to extinction.

Robodebt refers to the automated debt collection system that, between 2015 and 2019, wrongly accused hundreds of thousands of welfare recipients of overpayments.

Lis Ashby, head of policy and innovation at the Biodiversity Council, said Australia’s environment law – the Environmental Protection and Biodiversity Conservation Act – was “full of vague language and broad ministerial discretion”.

“Vague rules add to the current length of review processes because they prevent human reviewers from making rules-based decisions. The lack of clear rules will be even more problematic for an AI tool,” she said.

“Establishing clear rules in national environmental standards, including defining what is unacceptable, would speed up assessment times, even without the help of AI, and is important for any future adoption of AI. »

Brendan Sydes, national biodiversity policy advisor at the Australian Conservation Foundation, said the organization was “skeptical” of the Minerals Council’s efforts.

“Clearly, technology has a role to play in ensuring that nature conservation laws produce as effective nature conservation outcomes as possible. But while AI can be a good servant, it is a poor master,” he said.

He said the federal government should instead focus on the need to fill existing data gaps on endangered species and habitats.

Professor David Lindenmayer, a forest ecologist at the Australian National University and member of the Biodiversity Council, said research showed a third of Australia’s threatened species had not been monitored while others had only patchy data.

Evaluators filled in those gaps, he said, by consulting with experts.

“AI decisions are only as good as the data they are based on, and good data is not publicly available for most of Australia’s threatened species – often not even basic location data,” he said.

“AI automation risks making decisions based on faulty or outdated information, failing to protect biodiversity.”

Albania’s government passed reforms to environmental laws last year after a 2020 review found they failed to protect species and habitats.

Professor Hugh Possingham, a leading conservation biologist at the University of Queensland, said: “AI tools typically need hardware to train.

“The last 20 years of approval of the EPBC Act are clearly inappropriate, as the law has clearly failed to protect the environment. »

To speed up the assessments, he said the government should instead employ more people to carry them out.

Minerals Council chief executive Tania Constable said comparisons with robo debt were “disappointing” and the proposal was innovative and could strengthen environmental protection while improving efficiency.

She said: “The proposed approach would support human decision-making with AI tools for the regulator and project developer, including helping to navigate the complexity and variability of assessments and approvals under the EPBC Act.

A federal government spokesperson said budget decisions would be made “in due course”, but the Department of the Environment was considering how AI could facilitate applications.

“Decisions on whether or not to approve projects must and will always be made by assessment officers, not AI,” a statement said.

AI tools have the potential to save time, reduce uncertainty and translate technical language, the spokesperson added.