Have you fallen down the rabbit hole of AI-generated citations? It can seem like a bottomless pit. It’s our job as quality content producers to trace each citation to the original source and vet that source.
getty
There’s a moment I keep returning to. I’m looking at a list of citations in a commissioned research report that a client gave me—work created with the help of generative AI. I click one link. Then another. And another. And I keep falling down rabbit holes.
On the surface, the research I received looked robust. It had links to the sources and was technical in nature.
But as I dug, the floor kept falling away. Sources included blogs of uncertain origin and articles, even on reputable websites, that didn’t cite their sources.
Some of the source materials that did have attribution for their cases and data points provided references that were vague.
I knew that the commissioned research I received had been written by a machine, but how much of what the research pulled together was also written by a machine?
This is the paradox we now face.
The act of citation—once the reliable compass of journalists, academics, and credible creators—is simultaneously easier and more treacherous than ever.
With AI Citations, The Old Rules No Longer Apply
As a journalist trained to pursue editorial rigor, I was taught to follow the citation trail to its source. Editors drilled it into us: “Get the original. If you can’t, find a credible institution that has.” We had a healthy skepticism for statements that began with “sources say.” We looked for names, numbers, and the paper trail to the facts.
But now, AI can generate its own footnotes and citations, not to mention reports. We are flooded with references that seem solid.
But citations are no longer a guarantee of veracity. Instead, they are part of a new terrain we must navigate: Where does human knowledge end and machine-generated synthesis begin?
Even before generative AI, there was a creeping erosion of rigor.
I remember reading reports where a subject-matter expert (SME) at a consulting company would cite a figure to “industry sources.” Journalists would pick it up, and then it would appear in the press as “according to XYZ Consulting.” Suddenly, the number became legitimate by virtue of repetition, not investigation.
Now, magnify this with generative AI. We’re a few missteps away from becoming a society of parrots quoting parrots quoting machines.
Thought Leadership Demands Rigor With AI Citations
As someone who works in thought leadership, I feel this tension acutely. My clients rely on me to help them elevate their ideas—and to ensure their content stands on a foundation of facts.
That means doing the grunt work: tracing sources, verifying data, and reading original papers. It’s not glamorous. It’s often tedious. But it’s essential. Otherwise, thought leadership collapses into something without integrity.
The thought leadership world is actively working to address this and other challenges. The Global Thought Leadership Institute (GTLI), where I serve as a board member, is establishing standards for high-quality content—including sourcing practices. We understand that we need shared norms for quality and trustworthy thought leadership. And that’s what we’re working toward.
The New Psychology Behind Citations
There’s a cognitive shift going on around citations that we must acknowledge. Clicking a citation today no longer gives us an actual answer or source. It gives us options—often too many. You may discover something solid, but you may also be thrown into an infinite regress of interpretations, summaries, and machine-generated variants. Each one may be subtly different, causing you to think it’s a new or different source. Each one will beg the question: Where’s the actual original source?
We are left with citation fatigue, creeping distrust and the potential for intellectual paralysis.
Who has the time to vet every source in a 40-page deck created by generative AI?
And yet, how can we not?
In the past, the process of research was linear: read, evaluate, cite. Now, it’s uncertain. It feels like swimming in a pool where you’re never sure if the bottom is two feet down or two hundred.
Journalists Were Trained To Run Down Sources. Most People Weren’t.
This is where my journalistic roots are both a curse and a compass. I was trained to sniff out the weak link in a source chain. But many people writing for publication today weren’t.
They may be brilliant thinkers. But they haven’t necessarily been taught to distrust that seemingly credible post. Or to reverse-image search that graph or image. Or to ask: “Was this blog post paraphrased from a machine output?”
We are living through a disruption in knowledge at many levels.
The result is a proliferation of “insight” that isn’t insight at all. It’s echoes. We risk becoming monkeys quoting monkeys who are quoting monkey machines. And if we don’t act now—if we don’t insist on standards, on attribution, on intellectual integrity—we may never be able to climb back out of this hall of mirrors.
We Must Slow Down With AI Citations
I could have spent days more vetting the research that my client handed me. I almost did. But deadlines loomed. Expectations pressed. In the end, I made a decision: cite sparingly, use direct quotes where I could find them, and label uncertain sources clearly for future vetting.
This is what thought leaders must learn to do with their writing. In the era of generative AI, editorial rigor and standards must increase, not decrease.
We must slow down and reassert that great thought leadership is not fast content. It is written, constructed and edited with care. Brick by brick. Source by source.
This moment calls for ethical reflection: do we want to become content factories churning out gloss without grounding, where we quote each other until we forget what was ever fact in the first place?
Despite all this, I remain hopeful. If we, as thought leaders, recommit to quality, to transparency, to the sacredness of source-checking, we can elevate the entire field. The heart of thought leadership is integrity. And integrity shows up in the citations. That’s why we must ensure that citations are not used as decorations.
So here is my call to you: Be the one who checks. Be the one who follows the trail behind the AI citation. Be the one who says: “This isn’t good enough. Let’s dig deeper.”