I went to an AI prompt battle, and it was as bad as you think

There’s a special kind of silence that fills a room when enthusiasm meets reality. I felt it at an AI prompt battle last week during the largely enjoyable Upscale Conference. The competitive event rounded out a day of talks on how AI can be harnessed for creativity, where hopefuls typed words into text boxes while a crowd watched. What should have been an exciting, creative challenge was really like witnessing a live-action Google search.

The idea, I assume, was to celebrate the artistry of the pure prompt, those carefully crafted strings of words that conjure entire worlds from AI. Except, well, they didn’t. Most competitors stared at their screens, fingers hovering, trying to sound poetic while typing “fish-eye lens, girl in a room, fashion photo” or something, honestly, it was hard to tell. Every tap of the keyboard sapped the life from the room as the waiting audience hushed, murmured. and wondered, 'Is this it?'

AI has promised a lot: creative collaboration, a human–machine workflow to speed up old ways of doing things, maybe even the reinvention of art itself. But stripped of context, the 'prompt process' looks like someone shouting vague instructions at Photoshop and hoping for the best. Watching it unfold live was like peeking behind the magician’s curtain and discovering the trick is, well… a Theasauros, a spellchecker, and an encyclopedic memory for random culturally relevant things, or at the very least, a DVD collection to recall from.

“Guys, you had one thing to get right,” I found myself muttering in frustration as words appeared, were deleted, popped up again, fell off, and on, and on, and on. The energy was there, but the emotional connection wasn’t. Creating art, as it happens, is about more than words (just ask Gary Cherone and Nuno Bettencourt). Only one of three contestants managed to write and render a prompt in the time limit, and even then, the screen fell blank due to a technical hiccup. So, at the end, when the noise and buzz died down, there were no winners.

The AI prompt problem

And yet, this isn’t really their fault. Prompting, the act of coaxing AI into creativity with phrases, camera instructions, and style descriptions, has never been about the words alone. The best work I’ve seen in this space combines AI with real craftsmanship: Blender, 3D models, digital painting, and days in editing suites. The prompt becomes one piece in a larger puzzle.

Which is why 'prompt battles' feel so flat in 2025. After Coca-Cola’s recent AI ad campaign insisted there was 'more to it than prompts' but failed to convince, a focus on prompting alone falls into the trap of casting all AI use as the same flatly unartistic performance. I was hoping for a little mystery here, maybe collaboration, iteration, artistry. Instead, it was back to the old spectacle, reducing AI to its simplest form, and this, at the time, Freepik was launching Spaces, a node-canvas that connects AI use-cases, including prompts, video, and images.

With Spaces, as well as AI platforms like Flora and Firefly Boards, I can’t help but stay a little optimistic. Beneath the awkwardness, there’s a spark of something interesting in how AI can be used away from simply prompting and hoping for something to happen. The Prompt Battle reminded me how far AI art has come in just a couple of years, and how far it still has to go. Maybe next year, the 'battle' will move beyond the textbox. Maybe it’ll be about vision, collaboration, and merging old art with new, not just vocabulary and a large DVD collection.

If you do want to try prompting image and video creation, as it can be creative, visit Freepik for the latest AI models.