Deloitte used the Azure OpenAI GPT-4o, but said the corrections didn’t alter its findings

British multinational consultancy Deloitte will partially refund the Australian government after admitting it used generative AI to help produce a $440,000 report that contained multiple factual and citation errors.

Deloitte used the Azure OpenAI GPT-4o, but said the corrections didn’t alter its findings.

The report, commissioned last year by the Department of Employment and Workplace Relations (DEWR) reviewed the country’s welfare compliance framework.

University of Sydney academic Dr. Christopher Rudge, who first came across the errors, described the report as showing AI hallucinations. Labor senator Deborah O’Neill criticized Deloitte for relying too heavily on AI, calling it a “human intelligence problem.”

The issue highlights the negatives of overreliance on AI. However, it doesn’t seem to stop the technology from taking over. 

Deloitte recently announced a major deal to deploy Anthropic’s Claude AI to its 470,000 employees, signaling deeper integration despite the controversy.

Experts told The Wall Street Journal that AI could benefit consultancies in the next five years. 

In May, Anthropic CEO Dario Amodei said that AI could soon replace up to half of entry-level white-collar jobs.