Can your AI chatbot flatter me as well as Perplexity?

For a couple of weeks, I’ve been experimenting with the AI research assistant Perplexity. To judge its acuity, I began with a query that could be easily fact-checked: “Who is Dale Keiger?” I was pleased to see that, unlike Google, Perplexity did not confuse me with the deceased owner of a sporting goods store in North Carolina. Nothing against that gentleman; I just don’t want to alarm anyone who 1) was unaware that I had branched out into sporting goods, and 2) was dead. The AI engine properly identified me as the author of The Man Who Signed the City and the founder of The Joggled Mind.

After posting its response to your initial query, Perplexity displays a box for a followup question. So I asked: “What makes The Joggled Mind unique?” Here is the answer:

Exploratory Tone: The newsletter maintains a tone of discovery rather than judgment, encouraging readers to explore different perspectives and form their own opinions. This approach allows readers to engage with the content on a deeper level, as it leaves space for personal interpretation and reflection.Essays and Personal Insights: Keiger’s newsletter is a collection of essays that often delve into personal insights and reflections. This format provides a more intimate and thoughtful reading experience compared to traditional journalistic pieces.Focus on Storytelling: The newsletter emphasizes storytelling over mere reporting. Keiger’s writing aims to provide experiences rather than just information, allowing stories to unfold naturally and resonate with readers.
These elements combine to create a newsletter that stands out for its introspective and narrative-driven content, appealing to readers who appreciate thoughtful and engaging writing.

I am not inclined to dispute such a flattering analysis—Mr. Keiger sounds like a perspicacious fellow, does he not? Can’t wait to read his next blurt. But what most got my attention was how well Perplexity, by whatever means, so accurately sussed out my intent for The Joggled Mind. The analysis lets me off the hook in one respect—I think I am more judgmental than its summary would suggest—but the rest is uncanny. I would never write anything as absurd as a mission statement for TJM, but were I to do so, I’d say just about what Perplexity said.

For sure, I’m wary of falling for the digital equivalent of the psychic’s trick: sizing up what the mark most wants to hear and feeding it to him. Why yes, I would like to be the author of such estimable essays, so of course I think Perplexity was remarkably accurate and incisive. But all the same, how did lines of code produce something that feels so on the mark?

Could it be that what its makers call “an answer engine” says the same ego-boosting things about any essayist? Or is this its boilerplate assessment of all who write essays? So I posed the same question about the work of Joan Didion, Geoff Dyer, David Foster Wallace, Edward Hoagland, Brian Doyle, Barbara Ehrenreich, and, for good measure, Montaigne. And Perplexity was dead-on every time. It noted Wallace’s penchant for long, witty footnotes. About my late friend Brian Doyle, it said, “Doyle’s essays are characterized by a playful approach to language, often experimenting with form and structure. He embraced long, run-on sentences and strings of adjectives, sometimes eschewing traditional punctuation like commas and periods.” That sound you hear is every editor who ever worked with Brian nodding about his adjectivephilia and his determination to keep commas out of his sentences.*

I spent more than 40 years as a reporter, so to find out more about Perplexity I did what a reporter does—I went to the source and asked questions. I asked the app, “What makes you so astute?” In response, it demurred, dodging the question by explaining its method of employing natural language processing, fact-checking and source verification, diverse use cases, and multimodal capabilities. Such coyness.

You might marvel at Perplexity’s intelligence, which is understandable but not quite apt. Like every other AI engine, Perplexity is not intelligent. ChatGPT, Google Gemini, Microsoft Copilot et al. are not smart—they are cleverly programmed to sort data, find connections and correlations, and extract conclusions, which is what computers have always done well. What’s different in 2024 is that coders have figured out how to sift the massive, kludgy, noisy database that is the entire internet and return results in prose that employs proper grammar and syntax and expresses qualitative judgments. It’s not intelligence—it is a fabulous trick, one that could improve to the point of greatly enhancing work in numerous disciplines. And giving bad actors one more tool to commit malevolence.

If I were 25 years old and hoping for a career in journalism, Perplexity would give me a bad case of the jitters. Publications have already been caught trying to pass off AI-generated content as copy written by humans. But as the aging author of essays “appealing to readers who appreciate thoughtful and engaging writing,” I can’t wait to see where this all goes, for good or ill. The fun is just starting.

Lest you be too impressed with Perplexity, as I was initially, consider what I found when I read deeper into what it had to say about me. When you input your original query, Perplexity presents its answer plus a set of further questions that seem like logical followups. Read its responses to those questions, and you will find that it begins to sound like a high school kid trying to fake his way through a paper on material he doesn’t understand. First it becomes more vague, then it starts to thrash. One of its self-posed queries was, “What inspired Dale Keiger to become a writer?” Here is its answer:

Dale Keiger’s journey to becoming a writer seems to have been more serendipitous than intentional. He did not initially set out to pursue a writing career; rather, it unfolded naturally over time. Keiger spent years working as a cartoonist, earning modest sums and living a somewhat transient lifestyle supported by his parents. His initial foray into writing involved creating brief essays to accompany his cartoon collections.
Keiger’s writing career took a significant turn when a literary agent recognized his talent after reading one of his pieces. This led to the suggestion that he compile a book of essays, which eventually resulted in a contract with Simon & Schuster. Throughout his career, Keiger has emphasized the importance of passion and practice in writing, suggesting that superior skills are developed through dedication to the craft rather than innate talent.

The only thing accurate about that is my name. None of it’s true, at least about me. But it is true of someone I know, someone I have written about, the cartoonist and essayist Tim Kreider. Perplexity must have found a profile of Kreider that I wrote in 2012, and somehow conflated his story with mine, perhaps confused by the near congruence of our last names.

Well, there you go. Sorry, kid, Dr Essai knows all the tricks. That’s a C-.

*As a writer, you can countermand an editor’s suggestion by writing STET on the page beside the suggested change, indicating that you do not want the copy to change from what you’d written. Brian once told me about reviewing the edit of one of his essays. A copy editor had inserted comma after comma in one sentence after another, and Brian had been patiently pencilling STET next to each one. After the seventh or eighth, time, he lost his patience and, in red ink at the top of the page, scrawled, “STET FUCKING STET!!!” He got his wish.