Deloitte AI debacle underscores need for government contract transparency: OCP

Last month, Deloitte was required to partially refund the Australian government $440,000 after it admitted to using AI to help produce a report for the government, which ended up being full of mistakes, including made-up citations.

The errors were first picked up by welfare academic Chris Rudge from the University of Sydney, who identified over a dozen fabricated points attributed to his colleagues throughout the report.

Kaye Sklar, senior program manager for content and insights at the Open Contracting Partnership (OCP), said the Deloitte case had underscored the need for stronger AI procurement and contract management standards in government.

“As part of building their AI-readiness, public sector organizations should establish a clear vision for AI use cases, and when it’s appropriate - and not appropriate - to use AI. These expectations should be clear internally and with the vendors the organization works with,” Sklar told Accounting Times.

“This [Deloitte] case also demonstrates the importance of transparency around how and when AI is used, which is especially critical if AI is informing policy decisions.”

As AI usage becomes more widespread throughout professional services firms, Sklar said that governments would need to implement safeguards to maintain transparency, accountability and quality control throughout the contracting process.

In a bid to support transparency and accountability in government procurement, the OCP developed the Open Contracting Data Standard (OCDS), an international framework detailing how governments can publish data and documents at all stages of the contracting process.

“What open contracting data does is to provide a “public” for public procurement. That means that it’s open for everyone to see and review,” Sklar explained.

“It could mean that academics review the reports contracted out to consultants for accuracy, particularly as the attention is on this case.

“It also means that stakeholders within government can better learn from the case to ensure appropriate safeguards are put in place.”

Sklar added that governments could improve their guardrails and procedures surrounding AI procurement by adopting a strong national policy on AI.

“Chile first published a national policy on using AI in government in 2021. The policy puts a strong focus on the governance and ethics of implementing AI, alongside other practical considerations,” Sklar said.

Open data standards could boost transparency and accountability by enabling experts and the public to scrutinise government contractors during the procurement process, she noted.

“In AI, we are dealing with a field that is evolving incredibly fast, increasing the strain on a traditionally slow and bureaucratic function,” Sklar said.

“Governments need to respond by opening up their processes, including the data to ensure decisions are informed and data-driven, but also to enable better engagement with key actors inside and outside of government, to bring in the expertise where it is still being developed internally.”