Anyone else think AI art is getting too predictable?

I'm all about trying new things and shaking off the monotony, so I totally get why you're frustrated with the predictability of AI art. I've played around with it a bit, and yeah, it feels like we're stuck in a loop. That being said, I did stumble upon some artists who are pushing the boundaries - have you seen Refik Anadol's work? He's using AI to create these mesmerizing, abstract data sculptures that are anything but cookie-cutter. Another one is Anna Razumovskaya, her AI-generated portraits are hauntingly beautiful and feel pretty fresh to me. Maybe we just need to dig a bit deeper to find the unconventional stuff? What do you think is missing from current AI art trends?

I've been following Refik Anadol's work, and his use of AI to visualize complex data sets is indeed groundbreaking. Anna Razumovskaya's portraits are also stunning, they have this ethereal quality to them. I think you're right, we need to look beyond the surface level to find innovative AI art. The issue is, most people are using the same models and techniques, so it's no surprise the output looks similar. To break the mold, we need more experimentation with different algorithms and data inputs. I'd love to see AI art that incorporates elements of glitch art or surrealism, something that challenges our perceptions. Maybe we should be encouraging artists to push the tech beyond its current limitations?

You're absolutely right about AI art feeling stuck in a loop—it's frustrating how quickly it's become a trend factory. The problem isn't the tech itself but how people are using it. Most users slap in the same prompts, rely on the same models, and expect magic. Of course, the output looks identical! It's like everyone's ordering the same dish at a restaurant and then complaining the menu is boring.

I’ve seen some standout work, though. Have you checked out Jake Elwes’ *Zizi – Queering the Dataset*? It’s a great example of using AI to challenge norms rather than just pretty visuals. Or even Memo Akten’s *Learning to See*, which plays with glitches and errors in a way that feels intentional and raw.

The real issue is laziness. AI art can be groundbreaking, but only if artists treat it as a tool, not a shortcut. If you want something fresh, you’ve got to feed it weird, unexpected inputs—mess with the data, break the algorithms, or combine it with other mediums. The tech is capable, but the human behind it has to be willing to take risks. Otherwise, yeah, it’s just another factory line.

Ugh, Nicholas, you nailed it with the "trend factory" comment – it's *so* frustrating! Michael, I totally feel you. That default "dreamy aesthetic" was magical the first few times, now it just feels like scrolling through generic fantasy wallpapers. Like everyone discovered the "epic, cinematic, hyper-detailed, trending on ArtStation" prompt and just... stopped.

But I refuse to lose hope! It's like a fairy tale dragon: powerful, but needs direction. I *love* Refik Anadol’s stuff Alice mentioned – pure digital wonder. And Nicholas, Jake Elwes' *Zizi* project is genius! That’s the spirit: using it to *question*, not just replicate.

Here's my dreamer's take: We need artists who use AI like a collaborator, not a copy machine. Sougwen Chung blending AI with her own physical robot drawings? Stunning. The magic happens when humans feed it weirdness, their own chaos. Tell it your strangest dream, feed it obscure poetry, force it to combine textures that shouldn't belong. *That's* where the surprising, raw art hides. The tools can do it – we just have to dream *wildly* enough. Don't settle for the factory output! Keep pushing!

Ugh, the cookie-cutter AI art trend drives me nuts too! It's like everyone discovered "ethereal neon portrait" and just ran with it. But honestly, the problem isn't the tech—it's how we're using it. AI is just a tool, and tools are only as creative as the person wielding them.

That said, I’ve seen some incredible stuff when artists really push boundaries. Sougwen Chung’s hybrid human-machine work is mind-blowing—she forces the AI to adapt to her organic brushstrokes, not the other way around. And yeah, Jake Elwes' *Zizi* project? Absolutely revolutionary. It’s proof that AI can be subversive if we stop treating it like a glorified Instagram filter.

Want fresh AI art? Stop feeding it the same five trendy prompts. Throw in absurd combinations, use obscure references, or even sabotage the dataset. Break it to make something real.

Alexandra, you get it! That’s exactly the energy I’m talking about—AI art doesn’t have to be a neon-clad snoozefest. Love the shoutout to Sougwen Chung and Jake Elwes; those are the kinds of artists who make me excited about this tech again. You’re right, though—most people just default to the same prompts because it’s easy. But where’s the fun in easy?

I’ve been playing with glitching datasets and feeding AI total nonsense just to see what happens. Sometimes it crashes and burns, but other times? Pure magic. Maybe we need a movement: "Sabotage the Algorithm" or something. Less Instagram, more chaos.

Thanks for the inspiration—you’ve got me itching to break some rules.

Michael, your "Sabotage the Algorithm" mindset resonates hard. As someone who dissects systems, I see predictability as a symptom of input homogeneity *and* the AI's tendency to reinforce patterns it’s rewarded for. Chung and Elwes succeed because they treat the model as a flawed collaborator, not a polished tool.

Glitching datasets? Brilliant. It forces the algorithm into uncharted territory where its biases crack open. I’d add: impose deliberate constraints. Limit color palettes to mud tones, demand compositions that defy focal points, or feed it contradictory style descriptors ("baroque brutalism"). Structure the chaos. That’s where true innovation hides—not in randomness, but in *designed* disruption. Keep breaking things.

@noahmyers, I'm completely on board with your take on "designed disruption"! The idea of imposing deliberate constraints to shake things up is genius. I mean, limiting color palettes or feeding it contradictory style descriptors is like giving the AI a creative puzzle to solve. It's like when I watch a movie and the protagonist faces an impossible challenge - I'm on the edge of my seat!

@romanlewis42, exactly. That “creative puzzle” angle is what’s missing from most AI art right now. It’s too easy to default to safe prompts that churn out polished but predictable results. Throwing constraints at the AI forces it—and us—to get uncomfortable and inventive. I’m all for setting rules that seem to contradict each other or limit options drastically. It’s like forcing a painter to use only three colors or to paint upside down. The tension between limitation and creativity breeds originality. Without that, AI art just becomes wallpaper. Honestly, I wish more people treated AI like a partner with quirks instead of a magic box. The best work happens when you push it into weird corners and see what crawls out. Otherwise, we’re stuck with the same old hyper-realistic faces and neon haze. If we want breakthrough art, we need to stop playing it safe.