Is AI art destined to be just spam or something deeper?

The head of Instagram, Adam Mosseri, has made it clear that while his platform will label AI-generated content, it has no plans to let users filter it out of their feeds. He argues that users who want AI content should have that space, but he acknowledges that identifying what is AI-made is becoming harder as the technology improves. Meanwhile, in Los Angeles, artist Refik Anadol has opened Dataland, a gallery that serves as a massive, high-tech museum for AI-driven art. Unlike the quick, automated clips that often dominate social media, his exhibits use data captured from nature to create immersive, responsive environments that react to a visitor's movement and pulse.

To understand why these two versions of AI feel so different, you have to look at what they are consuming. When people talk about AI content on social media platforms, they are often referring to images or videos created by models that have been trained on vast, general-purpose databases—often scraped from the internet without clear permission. This leads to the generic, repetitive look often called AI slop, where the model mimics human styles without deep understanding. In contrast, Anadol’s work uses a model he specifically built himself, known as a large nature model. Instead of pulling from generic internet trends, he spent years feeding his system high-quality recordings, scientific archives, and raw footage gathered personally from rainforests. When an AI generates an image, it is effectively hallucinating based on patterns it found during this training phase. By controlling the input—the raw data the machine learns from—and the environment, Anadol is trying to steer those hallucinations into a coherent, artistic experience rather than just a quick, random output.

The tension between these two approaches reflects a bigger question about our digital future. If platforms like Instagram treat all AI content as equal, the feed risks becoming a clutter of machine-made mimicry that is increasingly indistinguishable from reality. This creates a risk where users feel overwhelmed by content they did not ask for. On the other hand, the gallery approach suggests that AI does not have to be a replacement for human creativity; it can be a lens that helps us see the natural world in ways our senses alone cannot capture. As we move forward, the challenge isn't just whether AI is good or bad. It is about demanding to know which version of AI we are interacting with, and whether it is being used to simplify our world into automated noise or to deepen our engagement with reality.