Scientists Warn AI Environmental Assessments Risk Robodebt-Style Failures
AI Environmental Assessments Risk Robodebt-Style Failures

Scientists Warn AI Environmental Assessments Risk Robodebt-Style Failures

Conservationists and scientists across Australia have issued a stark warning against a mining industry proposal to use artificial intelligence for environmental assessments, comparing the potential outcomes to the disastrous Robodebt scheme. The Minerals Council of Australia has requested $13 million in government funding to trial AI tools that would help companies prepare applications and assist federal regulators in decision-making processes.

The Robodebt Comparison

The Biodiversity Council, representing independent experts from eleven universities, told media outlets that while AI might assist with simple administrative tasks, automating environmental assessments "could lead to Robodebt-style failure, where computers make flawed decisions without transparency." Robodebt refers to the automated welfare debt-recovery system that between 2015 and 2019 wrongly accused hundreds of thousands of Australians of owing money to the government.

Lis Ashby, the Biodiversity Council's policy and innovation lead, explained that Australia's cornerstone environmental legislation—the Environment Protection and Biodiversity Conservation Act—contains "vague language and broad ministerial discretion" that already complicates human assessment processes. "The lack of clear rules will be even more problematic for an AI tool," Ashby emphasized, noting that establishing precise National Environmental Standards would accelerate assessments more effectively than artificial intelligence.

Wide Pickt banner — collaborative shopping lists app for Telegram, phone mockup with grocery list

Data Gaps and Species Monitoring

Professor David Lindenmayer, a forest ecologist at Australian National University and Biodiversity Council member, highlighted critical data deficiencies in Australia's conservation efforts. Research indicates that approximately one-third of the country's threatened species receive no monitoring whatsoever, while others have only patchy, incomplete data records. "AI decisions are only as good as the data they rely on," Lindenmayer stated, "and good data is not publicly available for most of Australia's threatened species—often not even basic location data."

Brendan Sydes, national biodiversity policy adviser at the Australian Conservation Foundation, expressed skepticism about the mining lobby's proposal. "While AI might be a good servant, it is a poor master," Sydes remarked, suggesting that the federal government should prioritize filling existing data gaps about threatened species and habitats rather than automating flawed processes.

Alternative Solutions and Government Response

Professor Hugh Possingham, a leading conservation biologist at the University of Queensland, argued that the government should increase human staffing for environmental assessments instead of investing in AI systems. "AI tools generally need material to be trained against," Possingham explained. "The past 20 years of EPBC Act approvals are clearly unsuitable material as the Act has demonstrably failed to protect the environment."

Tania Constable, chief executive of the Minerals Council, described the Robodebt comparisons as "disappointing" and defended the proposal as innovative technology that could strengthen environmental protections while improving regulatory efficiency. "The proposed approach would support human decision-making with AI tools for both the regulator and the project proponent," Constable stated.

A federal government spokesperson confirmed that budget decisions would be made "in due course" while the environment department explores how AI might simplify application processes. The spokesperson emphasized that "decisions about whether to approve projects must, and will, always be made by assessment officers, not by AI," acknowledging that AI tools could potentially save time, reduce uncertainty, and translate technical language in environmental documentation.

Pickt after-article banner — collaborative shopping lists app with family illustration