'Midjourney Magazine' đã có mặt– và nó là một cái gì đó không có ý nghĩa.

Midjourney Magazine has landed. The publication, a collection of thousands of AI-generated images as well as “interviews with Midjourney community members,” dropped its second issue this past week.

It’s a $4, 114-page coffee-table-style periodical filled with luscious, outlandish images and little else. There’s an eight-page interview, conducted by a human, with Bob Bonniol, a creative designer who has taken to using Midjourney to help iterate ideas. The Q&A aside, the title has very few of its own ideas at all.

The rest is just pages upon pages of large images of varying quality in varying genres, which are roughly grouped together based on theme and captioned with the prompt used to generate them, the human who gave that prompt, and the date they queried Midjourney, the generative AI platform from which its content is taken. (Midjourney is also the name of the company behind the tool, and the publisher of Midjourney.)

A profile photograph of a pensive female cyborg, all glossy, reflective metal skin, looking demurely down out of frame, sits on the same spread as what can only be described as a rejected character design for a Warhammer 40,000 figurine. The thing that seems to unite them in theme is “not human” and “a little disconcerting.”

Sometimes the themes slip. While the spread on pages 78 and 79 hangs together well, with an image of a woman in a yellow hat titled “in the style of fan ho, andy Goldsworthy, alex prager, anna atkins, franco fontana, Rosalyn drelxer ::1 umbrellas ::-0.02 styled by alan lee ::-0.48” sitting alongside “don’t look at the eloquent red circle, surreal, glistening highlights –ar2:3 –s 33” (which features a woman looking at a blood-red moon), others don’t do so well. Three cats wearing bathrobes doing tai chi in a bonsai-filled courtyard sit on the page opposite an image of a man walking on the Big Apple’s sidewalk that wouldn’t look out of place on Humans of New York.

It is enormously impressive to flick through. But when you start to look for more it falls flat. “It looks like a standard glossy magazine, with nice pictures and a simple layout,” says Michelle Pegg, cofounder of Curate Creative, a UK-based creative agency, “but as a magazine is a vehicle for stories and expression, and connecting with the reader, I feel it goes no further than the set of glossy pics.”

And I agree. In part this could be sour grapes: I work in an industry that has historically relied on being able to sell magazines as a luxury product, one carefully curated for you. They’re so expensive because, as advertising revenues that subsidized many titles have disappeared, publishers have been loath to scrimp on standards. Photo editors cost money. Designers, too. Journalists and editors and fact-checkers don’t come cheap.

But AI does—at least it does when its huge computing costs are subsidized by venture capital or the beneficence of Big Tech firms, as has happened so far with the rise of generative AI.

Yet the things that cost money are the things that give magazines their quality. The ability to see something you hadn’t expected is what separates printed products from the internet. It’s why those who love magazines do so fiercely. And it’s why I’m conflicted by Midjourney Magazine. I want to like it. But it’s soulless.

Pegg explains the problem well, comparing it to the “alt text on a website image.” Her main problem with the magazine echoes the fear many have with AI writ large. “The big thing missing in the magazine is the human connection,” she says. “No stories, no obvious reason behind the images that I want to know more about, no reason for that style.”

She says that the magazine has “no depth, just pretty enough pictures.” And she has qualms—as many do with AI-generated images—about the extent to which the work skirts on the right side of copyright laws.

One photo, on page 11, shows the results of a prompt asking for a 1940s-style photograph of a woman looking like Judy Garland, which almost exactly matches her facial features, suggesting the underlying model has been trained on images of the Hollywood icon.

“How will we know if what’s produced isn’t plagiarizing an artist’s work as it draws from what’s already out there?” Pegg asks. It’s an issue that Midjourney is reckoning with right now—it’s currently facing a class-action lawsuit over alleged copyright infringement. Midjourney claims in its defense that none of the plaintiffs in the case can point to their art being used as training data.

The magazine’s tagline ends with the claim it is “expanding the imaginative powers of the human species.” That’s something Pegg doesn’t necessarily dispute—some people will feel that way, she’s sure—but she does admit it doesn’t feel that way for her. “My first question on anything is always, why? What’s behind this? What’s the story?”

That’s not a question AI feels well equipped to answer—at least not yet.