Once, there existed an era where one could comfortably assume that an op-ed in a major publication was penned by a human being. Cherish that thought as long as possible.
This week, billionaire investor Stanley Druckenmiller stirred discussion after publishing a critical piece in The Wall Street Journal aimed at Treasury Secretary Scott Bessent, a former mentee of his. Observant readers noted patterns that suggested the piece was generated by artificial intelligence. The AI detection platform, Pangram, confirmed that the op-ed was entirely AI-written.
While this revelation itself wasn't ground-shaking—AI-generated content has made its way into mainstream media, including op-ed sections, despite intentional prohibitions—the response was notably different this time. Generally, those accused of misappropriating AI-generated text tend to apologize or refute the claims. For instance, in April, James Taranto, the Journal’s editorial features editor, referred to Pangram as a “defamation machine,” arguing that it unfairly targeted op-ed writers he oversaw. He emphasized the ethical duty of professional writers to ensure originality in their work.
Druckenmiller, however, took a different approach. When questioned by NOTUS, he openly stated, “Of course I used AI. I’m not embarrassed by it. I use AI for everything now, just like I use a calculator for math.” Instead of criticizing him, Paul Gigot, the editorial-page editor of The Wall Street Journal, backed his decision. “AI is a fact of modern life,” he declared. Gigot later elaborated in an entire column, ridiculing the “media pharisees” who chastise the use of automated writing. He asserted that while the Journal's opinion writers must produce their own drafts, he has no intentions of monitoring AI usage among external contributors. “Prominent politicians and CEOs often resort to speechwriters or writing firms,” he pointed out. “How is that different from using AI to enhance a piece?”
Druckenmiller’s nonchalance toward AI usage reflects a broader trend among some in the business and tech sectors who view AI as an invaluable tool for various tasks, including writing. Venture capitalist Chamath Palihapitiya echoed this sentiment on X, stating that if Druckenmiller isn’t ashamed, neither should anyone else be. He likened clinging to traditional human writing practices to celebrating the laborious process of starting a fire with sticks after the invention of matches. Yet, a prestigious newspaper like the Journal officially embracing AI is an unprecedented move. Could this imply a shift toward a stigma-free acceptance of AI-generated writing?
Gigot’s perspective is certainly intriguing, especially in a landscape where many writers remain reluctant to acknowledge AI’s influence. His claim that an AI-written piece can still encapsulate a writer's genuine ideas is quite plausible. While I'm not equipped to evaluate the specifics of Druckenmiller's op-ed within the context of the bond market, it does seem to express a significant viewpoint that mirrors his thoughts, irrespective of its authorship. After all, op-eds generally aren't known for showcasing literary brilliance.
However, the notion that Druckenmiller and Gigot deserve accolades for their candor overlooks a critical issue: transparency. Until this moment, there was a consensus among those engaged in discussions about AI writing about the necessity of disclosure. The opinion section of the Journal appears to have sidestepped this convention. Druckenmiller’s piece offers no hint about AI involvement, and Gigot's response dismisses reader concerns about needing to know if a piece utilized chatbots. “Disclose what?” he rhetorically questioned. “Should we question an author on how they used ChatGPT? Some publications claim to do this, but it often feels more like a formality than an actual process.” (Gigot has not responded to my inquiries, nor has Dow Jones made any comments on this issue. Taranto reiterated his support for his earlier statements.)
If the Journal’s leaders genuinely believe that readers are indifferent to whether an op-ed was AI-generated, then what prevents them from revealing such information? Research suggests that while readers sometimes prefer AI-written narratives, their enjoyment diminishes once they learn about the AI's role. Gigot may interpret this as a sign that people appreciate AI's contributions without realizing it, but another possibility is that readers prefer engaging with text produced by human intellect.
Outsourcing an op-ed to an AI could be compared to engaging a ghostwriter, but this analogy urges a re-examination of ghostwriting practices instead of merely defending AI. The Journal's transition from denying AI influence to embracing it as a matter-of-fact feels premature. Although many disruptive technologies eventually gain acceptance, resistance often serves as a crucial way for society to refine norms, establish boundaries, and allow time for adaptation. Presenting AI-generated content as human-authored without disclosure may ultimately undermine trust in the technology rather than solidify it.
Another compelling reason for maintaining the standard of disclosure revolves around the writer's awareness of their own thought processes. When individuals allow a chatbot to assist in writing, even if they are providing guidance, they may not recognize how profoundly AI can influence their thinking. Jenna Russell, a computer science doctoral candidate at the University of Maryland, highlighted a significant finding from her research into AI writing: AI-generated works often fail to introduce novel ideas compared to fully human-written texts. Russell explained, “These large language models can start shaping our beliefs without our explicit realization.”
Druckenmiller's embrace of AI, coupled with the Journal's endorsement, may signal the dawn of a new era for automated writing without guilt. Alternatively, it might exemplify a troubling trend in which societal elite escape consequences for actions that others would not. In contrast to Druckenmiller's privileged acceptance, the treatment of emerging writers, such as Mia Ballard—whose AI-assisted horror novel faced backlash and cancellation—illustrates a jarring disparity.
The eventual trajectory for The Wall Street Journal and the broader media landscape will likely hinge on market dynamics. Should there be sufficient audience demand for chatbot-generated opinions, other publications may readily follow suit, capitalizing on a stream of eager contributors ready to transform AI prompts into published content.
