Adobe’s Generative AI is Half Baked at Best
I’m no stranger to Adobe products like Photoshop, Illustrator, and InDesign. I’ve been knee-deep in graphic design projects for my various electives these past few months. When I noticed I had received the update to Adobe’s Creative Suite that includes generative AI, I was quick to test its limits. (My experimentation with AI art was for personal use only and was not submitted as classwork.)
In this post, we’ll dive into an amateur graphic designer’s honest opinion on whether or not these new features are genuinely usable.
Adobe Photoshop
Photoshop’s Generative AI is marketed to have the ability to fill, edit, and remove image elements based on a written prompt.
I started with a base photo. My beautiful friends and I from our holiday party. My goal was to create an image using three AI elements: Remove a subject, add a subject, and generative expand to create a larger image.
I sought to remove my friend Maura on the right of the image, replace them with Santa Claus, and expand the picture so you could see more of the host’s apartment. Here’s what happened.
Feel free to zoom in on the hands if you’re in the mood for nightmares.
I’d say the AI element removal feature is pretty impressive. No trace of Maura! I could see a photographer finding a lot of joy in removing unwanted elements of an otherwise perfect photo with just the click of a button. I believe Photoshop AI struggles with humans. I think the image speaks for itself. The hands. The Santa. They wouldn’t pass as fit for a Christmas card, but they’d do well in a horror film.
Photoshop AI is functional at best but not to be used on a large scale for any project you care about. The object removal feature is your best bet when using AI on this software.
Adobe Illustrator
I asked the AI in Illustrator to create three forms of images from scratch using only words as a prompt. The first was a scene of a beach alongside the city. I think the illustration concept is immediately identifiable and executed quite well, but a close look would reveal that the image was AI generated. Certain shapes just look a little wonky.
Creating subjects seems more like Illustrator AI’s strong suit than scene creation. Two robots in a similar style were created, and a matching pattern complemented them. In terms of realistic usability, a brand could easily make a quick graphic using this feature. If anything, it made a better vector robot than I ever could.
Illustrator’s AI functions hold their own despite some of its capabilities being a little half baked.
In general, I do find these tools to be helpful in small doses. Overall, humans will always create better art. Our understanding of the world is one an algorithm will never be able to truly grasp (and I hope it never does). As far as Adobe’s program’s go, the addition of AI is certainly impressive, but only when used in context with already impressive human work. It’s a party trick, at most.
My reccomendation? Give it a try, but learn how to do it yourself first.
