🎙️ Voice is AI-generated. Inconsistencies may occur.
"Why haven't I met Tommy?"
That's what my girlfriend wanted to know, as she forwarded me a weblink.
The article in question—a biographical recounting of my life—revealed a litany of fascinating facts. I'm a "world class" singer who has opened for folk-rocker Joan Osborne in Nashville. I'm a passionate adherent of Transcendental Meditation, which I was introduced to by my mentor, the filmmaker David Lynch. I'm a former executive director of the Connecticut Humanities Council.
And—most salient to my girlfriend—I'm also apparently the mother of a 12-year-old son who likes basketball and comedy (and whom she'd never met).
Of course, none of it was true.
Artificial intelligence—and its potential ability to destroy humanity—has been in the headlines for years, accelerated by the breakthrough introduction of ChatGPT last November. World domination may eventually be the on the horizon, but for now, the more immediate AI-based threat is both banal and cuttingly personal: SEO link farms are now using generative AI to create (invented) content at scale, attempting to make money by driving clicks. And sometimes that content is about...people like me. Or, perhaps, you.
Since the earliest days of the internet, poorly paid "article spinners" or "content spinners" have beenhired by shady companies to create articles—much of them recycled or stolen from other sites, and sloppily reassembled—for the express purpose of driving web traffic to the links they embed. But actual humans are becoming less necessary as generative AI exponentially ratches up the speed at which content, whatever its level of accuracy, can be created.
Indeed, the article chronicling my life—which was recently taken down, upon my request—seemed to be a classic "content farm" piece, touching on keywords that Google has identified as commonly associated with me, and filled with links to not-exactly-germane websites focused on the comparative virtues of various HVAC systems and the exchange rate of the Singaporean dollar. It also bears the creative marks of AI, which seems to have picked up the "gist" of my life without much specific concern for the facts. Indeed, the site elsewhere notes that "A portion of our articles are written by AI under human supervision. We fact-check the articles and update them periodically." (Though perhaps not very well, as the site's contact page directly links to a complaint form to report inaccuracies and demand content removal.)
The article had some things down correctly: my profession as a business author and speaker, my alma mater. It knows I have something to do with the arts and humanities (I write musical theater and have a graduate degree in theology), but its grasp is vague (thus, opening for Joan Osborne and running the Connecticut nonprofit). It correctly notes that I've directed a documentary film, but has decided it's about the future of the gaming industry. Not terrible guesses, but also untrue.
Most strangely, and egregiously, was my 12-year-old, Tommy. Because of a popular TED.com talk I'd given in which I reference the death of my cat, the search term "Dorie Clark son" became popular—and apparently, AI took the ball and ran with it, creating a "a talented and witty young man who shows exceptional promise both on and off the basketball court." Rather impressively for a 12-year-old, the site reports that "In his recent game against a fierce rival, Tommy used a trick he learned from reading one of his mother's books on leadership and teamwork, and led his team to victory." I've never wanted kids, but—sorry, bragging is a mother's prerogative—Tommy sounds pretty terrific.
The presence of the article was a fairly minor inconvenience. It's unlikely, though not impossible, that someone might have been fooled by the incorrect information. Though if it were picked up on other sites—or webcrawled, as fact, into future versions of AI—it might ultimately become a problem.
But the real concern is that I'm a civilian, not a celebrity. My life isn't anonymous, but it's a far cry from Taylor Swift or the Kardashians. When human labor was required to create random, clickbait articles, there were limits on how much could be created. It simply wasn't worth it to write lengthy, detailed articles about private citizens in hopes of generating a few extra clicks for an HVAC comparison site. But now, thanks to AI, the marginal cost is zero. There's simply no economic reason not to flood the internet with junk articles, replete with invented facts about any of us.
I'm lucky AI decided to paint a flattering picture. (Though it inflated my age, it also authoritatively declared, "some say that she looks even younger than people in their 30s.") But we might not all be so fortunate.
The existential threat of AI—according to everyone from Elon Musk to thousands of high-level researchers—may indeed be real. But for the near-term, the fundamental challenge is what it's always been with new technologies: hucksters leveraging a new angle to make more money, regardless of the consequences to society or the individuals they use to generate more clicks. Tommy would be appalled.
Dorie Clark teaches executive education at Columbia Business School and her latest book is The Long Game: How to be a Long-Term Thinker in a Short-Term World (Harvard Business Review Press). You can download her free Long Game strategic thinking self-assessment at dorieclark.com/thelonggame.
