Article · AI translation
IT Home reported on November 8 that Accenture has been accused of publishing a research report riddled with AI “hallucinations,” containing not only a large amount of false information but also several apparently fabricated research cases.
According to a recent analysis by GPTZero, a report published by the consulting giant allegedly contains 15 false citations and entirely fabricated case studies, and is therefore facing outside scrutiny.
The report, titled “Leapfrogging Development,” was published jointly by Accenture and researchers at IE University in Madrid. It aims to explore how AI can improve productivity in Africa’s agricultural sector.
Accenture denies that the errors in question were caused by AI.
Nevertheless, the incident could still leave Accenture in an embarrassing position. The company’s management previously warned that employees unwilling to use AI could face dismissal.
Accenture CEO Julie Sweet warned last year that, as part of the company’s AI transformation, consultants who failed to use the technology would be “phased out” by the company.
GPTZero’s report highlights concerns arising from the increasingly widespread use of AI in the consulting industry. One risk is that AI “hallucinations” could spread false information through chatbot answers and summaries, thereby “contaminating information sources.”
GPTZero said: “The company believes fabricated citations are an immediate, real threat to researchers, academics, consultants, students, and anyone who searches for information on the internet.”
This is not the first such incident in the consulting industry. Major consulting firms including Deloitte, EY, KPMG and PwC have previously been accused of including AI errors in their reports.
Meanwhile, some of the world’s largest and most expensive law firms have also acknowledged AI-generated errors in important legal documents.
Pinsent Masons, one of the UK’s major law firms, previously acknowledged that legal documents it submitted to the UK High Court contained AI “hallucinations.”
Lawyers at another firm, Sullivan & Cromwell, charge up to $3,000 an hour (IT Home note: approximately 20,058 yuan at current exchange rates). The firm also acknowledged earlier this year that documents submitted to a New York court contained AI errors.
Accenture is among the many professional services firms actively encouraging employees to use AI.
However, demand from employees was so strong that the company was forced to impose new restrictions on the use of AI tools last month. Some employees had previously run up monthly bills of more than $1,000 (approximately 6,686 yuan at current exchange rates) using tools such as Claude.
An Accenture spokesperson disputed GPTZero’s findings, saying: “Our review indicates that the errors found in the report resulted from human error. There is no evidence to support the conclusion that these errors were caused by AI hallucinations. We are correcting the report and will complete that work before republishing it. We always welcome opportunities to further improve the quality of our reports.”