By Joel Stein
When AI steals all the jobs, it's clear what will be left: celebrities. Also, welders, roofers, mechanics, home health aides, masons and beekeepers. But celebrity seems more attainable for me.
So when I opened my favorite academic journal for marketing, the Journal for Marketing Research (no offense, Journal of Marketing), and saw a paper titled, "An AI Method to Score Celebrity Visual Potential," I knew I had to get my celebrity visual potential checked. If it went well, I could become an influencer and avoid the sting of getting rejected, or accepted, by beekeeper school.
The paper's authors, academics at Carnegie Mellon's business school, stripped photos from the internet of 6,000 celebrities and 6,000 nobodies. The celebrities were taken from entertainment, business, politics, social media and sports -- but oddly excluded contributors to The Wall Street Journal.
They fed the photos to an AI algorithm, asking it to discover the facial characteristics that separated the two groups. They set the minimum score for a celebrity at 0.5. Anything below cut off and you are doomed to a real job and as few as one spouse for your entire life.
This was the kind of experiment with massive amounts of data that was impossible before AI. And the results were insanely accurate. When they fed their trained algorithm a photo of a random person, it was 96% likely to correctly guess if that person was famous. This goes for all celebrities, even those with jobs seemingly less reliant on looks, such as football players, senators and theater actors.
While Halle Berry and Elon Musk might have very different faces, both are high in charisma, and the paper argues that's largely because they seem warm, trustworthy, friendly, nonaggressive, competent and dominant. Because the judge was a computer program, it couldn't be swayed by recognition. Instead, it seemed to be weighing the characteristics that are typical in all celebrities, such as high cheekbones and big eyes. It's the kind of discovery that will further shift Hollywood capital from publicists to plastic surgeons.
As soon as the authors agreed to assess my face for its celebrity potential, I sent them six photos. They weren't my best photos, but they were very close to my best photos.
Two days later, I got an email from the lead writer on the paper, Xiaohang Feng, a Carnegie Mellon Ph.D. student. I paused before opening it, more nervous about my results than I was about my skin cancer biopsy. Skin cancer, after all, is curable.
Feng wrote, "Your Celebrity Visual Potential (CVP) score is impressively high, exceeding 0.8. Your high CVP score is mainly boosted by your prominent cheekbones, large eyes and narrow facial width-to-height ratio, but reduced by your relatively long mouth-nose distance. According to past psychological literature, these facial features make you look more trustworthy, warm, competent and nonaggressive, and thus charismatic."
I read it again. "Relatively long mouth-nose distance"? I went to a mirror to examine my philtrum, and then did so much panicked internet research that I learned the word "philtrum." It was indeed Frankenstein's monster long. Though Frankenstein's monster is a celebrity.
The problem with a big philtrum, the journal paper states, is that "researchers have found that a longer mouth-nose distance may predict sarcasm." Yeah, I thought, right.
After throwing out the lowest score and the highest score, which is the scientific procedure you use when you don't like one of the results, I averaged to an 0.889. To put that number in perspective, my mug is apparently more charismatic than John F. Kennedy's (0.708) or Beyoncé's (0.700 -- yes, really). Barack Obama has an 0.860. While I am no Michael Jordan (0.999) or Pope Francis (0.920), who's apparently skating by on his looks, perhaps I could have ended apartheid faster than Nelson Mandela (0.701).
Feng, the study's author, and her adviser, Professor Kannan Srinivasan, were all too clear in their paper -- and again, insistently, on a Zoom call -- that a high Celebrity Visual Potential didn't correlate with being good-looking. In fact, they admitted they were surprised by just how unattractive most celebrities are. They found that the correlation between fame and conventionally attractive features is only 17%.
It was this discrepancy between fame and beauty that first made Feng interested in the topic. "I was thinking about what makes a celebrity charismatic. People previously thought it was only attractiveness. But some actors and actresses aren't good looking," she said. I nodded at the screen while holding a finger to my mouth in a way that both showed deep understanding and hid my philtrum.
A persuasive face like mine, they said, has the following characteristics, ranked by importance: short width-to-height (narrower faces appear more trustworthy), sexual dimorphism (science, unlike humans, says I look manly), averageness (nothing too weird), high cheekbones, slightly dark skin (Feng -- and not me -- says past psychological studies -- and again, not me -- theorize that darker skin is associated with dominance, which correlates positively with charisma, but also aggressiveness, which correlates negatively), a large mouth-to-chin distance (dominance, competence), big childlike eyes, symmetry, a short philtrum (whatever), and not looking like a baby (babies are rarely famous).
Srinivasan wasn't surprised that I did so well. "The moment I saw you, I thought you were charismatic," he said, though I knew it was the kind of thing you had to say when you've discovered that the person you're talking to has a CVP score of 0.889. Still, he said, my face could only take me so far. "I don't want you to quit your job and start competing against Tom Cruise." Of course not, I assured him. Though I could probably steal some roles from Julie Andrews (0.8489).
Yet there were a lot of things that they thought I could do because of my charisma. Sales, public speaking, television hosting, social-media influencing. Feng thought I could accomplish even bigger things. She guessed that by hiding my problematic philtrum beneath a mustache or a touch of lipstick, I might break the 0.9 barrier.
I was feeling pretty good until I remembered that I've had this face my whole life. Still somehow I've lost every TV hosting job I've auditioned for, never been invited back to a speaking gig, failed to sell my last four TV pilots and have been married to the same woman my whole life.
I'm a huge disappointment to my face. It's being held back by my lazy, loyal, untalented ass. Which, luckily, no one has figured out how to give a CVP score to yet.
Joel Stein, a former columnist for Time, writes a Substack called "The End of My Career" and is the author of "In Defense of Elitism."
(END) Dow Jones Newswires
April 18, 2025 09:30 ET (13:30 GMT)
Copyright (c) 2025 Dow Jones & Company, Inc.
Disclaimer: Investing carries risk. This is not financial advice. The above content should not be regarded as an offer, recommendation, or solicitation on acquiring or disposing of any financial products, any associated discussions, comments, or posts by author or other users should not be considered as such either. It is solely for general information purpose only, which does not consider your own investment objectives, financial situations or needs. TTM assumes no responsibility or warranty for the accuracy and completeness of the information, investors should do their own research and may seek professional advice before investing.