PwC is the most recent Big Four firm to be found releasing low-quality AI content.
The largest consulting firms in the world are offering costly guidance to companies on adopting AI while avoiding pitfalls. However, their own reports continue to face scrutiny. PwC has recently come under fire, with four of its “thought leadership” reports in the Middle East being described as problematic by the Financial Times. These reports on AI and electric vehicles included erroneous footnotes, misattributed statements, and fabricated references. The research group GPTZero identified these inaccuracies as AI-induced errors, and the FT confirmed them.
One footnote ended with ‘chatgpt.com.’ The issues were quite apparent. In one report, a PwC survey claimed that 70% of chief executives in the Middle East believe generative AI will transform their businesses. However, the footnote linked to a news article that did not mention this survey. Another footnote included a web address that still had the tag “utm_source=chatgpt.com.”
“The messy attribution of source material reflects the nature of AI-generated research,” stated Paul Esau from GPTZero to the FT. This firm promotes itself as an advisor on responsible AI use, emphasizing how to avoid such mistakes.
PwC is not the only offender in this situation. It has now joined a negative trend. GPTZero's earlier findings prompted KPMG to retract an October report that contained inaccuracies regarding AI practices at UBS, the UK’s NHS, and Transport for London. Last month, EY withdrew a study due to false footnotes, and Deloitte has inserted AI inaccuracies in government reports twice, including a $1.6 million health plan in Canada that referenced nonexistent papers.
The same pattern emerges each time. The Big Four have produced countless AI thought-leadership articles to attract clients while encouraging their employees to leverage AI for efficiency. The outcome is a collection of subpar content branded with prestigious logos.
The timing of these revelations is almost too coincidental. In the same week, PwC’s US CEO Paul Griggs was discussing with Business Insider where companies typically misstep with AI. One pitfall he mentioned was implementing AI on dysfunctional processes. “All AI will do is highlight how poor your messy process really is,” Griggs remarked. His own firm just validated this point.
The uncomfortable implication is what this indicates about the quality of their output. These firms command high fees for their expertise and thoroughness; still, it appears that no one verified the authenticity of the sources. This gap mirrors the issue seen in AI-generated code that may contain security vulnerabilities: the output appears correct, so scrutiny is neglected.
Moreover, this runs counter to their promotional claims. Consultants are among the most vocal advocates telling clients that AI will revolutionize their operations. Ironically, their reports illustrate how it can subtly undermine them and show how the enthusiasm for AI often exceeds safe application. The inaccuracies are a clear indicator, yet the detection continues to uncover them. Nevertheless, the firms persist in publishing their content.
Other articles
Anthropic remains alone in every AI battle, yet emerges victorious regardless.
Anthropic stands out as the outlier in almost every AI policy debate, ranging from open weights to the Pentagon. Moreover, it remains the most valuable startup globally.
Shein reveals an FTC investigation as it seeks to go public.
Shein revealed an unspecified FTC consumer protection investigation in its IPO filing in Hong Kong. The agency regulates 'dark patterns,' which are the manipulative tactics used by Shein's app.
Cyera acquires Oasis Security for $1 billion to monitor AI agents.
Cyera is acquiring Oasis Security for approximately $1 billion to manage the "non-human identities" that AI agents use to access company data, marking the largest security deal in the AI-agent sector to date.
As AI produces an increasing number of messages, the ability for humans to recognize them becomes more valuable.
AI has accelerated internal communication, but it has also made it more difficult to trust. Motivosity suggests that integrating company updates into recognition platforms, where employees are already engaged, is the way for organizations to stand out amidst the clutter.
Apple's new Upgrade plan won't impose restrictions for missed payments; however, three missed payments could terminate your lease.
Apple states that missed payments under its Klarna-supported Upgrade plan will not affect app access or device functionalities; however, three consecutive missed payments may lead to the termination of the lease.
BuzzFeed lays off a third of its workforce, and AI was not the solution.
BuzzFeed eliminated approximately 180 positions, which is a third of its workforce, during owner Byron Allen’s initial major layoffs. Its investment in AI has not stopped the downward trend.
PwC is the most recent Big Four firm to be found releasing low-quality AI content.
PwC released Middle East reports containing numerous fictitious footnotes and invented sources, with one URL ending in “chatgpt.com.” This marks the fourth instance of a Big Four firm being caught using subpar AI content.