MidJourney offers a feature where users can upload a photo and receive a detailed description generated by AI. Originally designed for accessibility and image enhancement purposes, it can also be used for fun, such as creating alter egos.
Recently upgraded for enhanced accuracy and descriptiveness, I decided to test it on my colleagues at Tom’s Guide to see how well it could analyze their images.
Starting with myself, the AI described me as “A photo of an overweight man with glasses and a goatee, sporting an orange trucker hat labeled ‘Nashville’ and a blue t-shirt with ‘Grom’. The image, resembling the work of an unknown artist, was posted on Reddit in 2018.”
Moving on to Ryan Epps, a tech news writer, the AI humorously labeled him as “an overweight man with brown hair and a beard, taking a selfie in a grey hoodie with ‘dagnafkin’ written on it. The photo, resembling an anonymous user’s style, was posted on Reddit in 2019.”
Hunter Fenollol, a smart home editor, received a flattering description as “an attractive man with dark hair and a beard, wearing a necklace and a black tank top in his suburban New Jersey driveway on a sunny spring day. The image was likened to realistic portraits.”
Alex Wawro, the computing editor, was amusingly compared to Eddie Redmayne in the AI’s description: “resembling Newt Scamander from Harry Potter, wearing a brown leather jacket and a black shirt, giving a thumbs up while enjoying tacos and beer at an outdoor bar at night.”
Cynthia Lawrence, the homes content editor, was recognized as “an attractive African American woman with long black hair, sitting in her home office with red and white blinds, wearing a blue cardigan over a gray t-shirt and pink lipstick. The detailed description impressed her despite the visual disparity.”
Mike Prospero, the U.S. Editor in Chief, was mistakenly identified as a software engineer in the AI’s analysis: “depicted as a male software engineer with short dark hair and a beard, dressed in a blue gingham shirt at Bryant Park in New York City during springtime.”
Nick Pino, the managing editor for TV and AV, was praised as “an attractive young man with glasses and short hair, standing on the Golden Gate Bridge in San Francisco, smiling at the camera. Despite the visual mismatch, he appreciated the AI’s effort.”
Kate Kozuch, the Managing Editor for social, was described as “an attractive young woman with long blonde hair, sitting in the passenger seat of her car, smiling at the camera. The AI suggested the image was posted on Snapchat in 2018, leaving Kate intrigued by the voluminous hairstyle reference.”
The AI’s interpretations sparked varied reactions from my colleagues, ranging from amusement to surprise, showcasing the fun and unpredictable nature of AI-generated descriptions.
