Individual Expression or Machine Output: Why We Need to Start Labeling AI Art Now

In late 2023, a few of my friends started experimenting with AI art generators online. Curious, I joined them, typing prompts into the generator and seeing what it would create. That first day, I asked the website to make a watercolor painting of a hummingbird, and seconds later, a vibrant little bird appeared on my screen. Pretty, I thought. Then I closed the tab and returned to my sketchbook, where I’d been working on a series of owls, some mid-flight, others perched onto a tree branch. What I was drawing by hand wasn’t necessarily higher quality than the AI-generated images. But while one had required almost nothing from me, the other reflected something about who I was, in the thousands of tiny decisions I made in how I drew the things I wanted to draw.

Art, to me, is a way of expressing and sharing our individual consciousness. Every human-made drawing carries parts of the artist behind it: the choices they make, the time they spend, and the experiences they bring, all playing a part in the final piece. AI, meanwhile, can imitate styles and produce beautiful images, but it cannot express a life it has never lived. Yet AI art is everywhere today, from our social media feeds to the major auction houses. And while most people still value human-made art more, the problem is that we’re actually quite bad at telling it apart from AI-generated work. People deserve to know whether the art they are looking at comes from a human or a machine—whether it’s an expression of an individual consciousness, or output from a machine. That’s why it’s time we make a concerted effort at labeling AI art as such.

Art has been part of my life for longer than I can remember, and over the years, it has become much more than just a hobby. I’ve given drawings as birthday gifts to family members and met close friends through fan art. This year, I drew my mom a sunflower for her birthday because sunflowers have always symbolized our relationship to me. She is the sun, and I am the sunflower: I grow in the warmth of her love and guidance, always turning toward her light. It wasn’t perfect, but every decision, from colors to shading, reflected my own effort and feelings toward her. 

Critics may argue that if an image looks beautiful, why should its origin matter? If I had simply generated an image with AI and printed it out, my mom would have still thanked me—but she would have known that I hadn’t invested the same time or thought into making something for her. That’s because art is more than a finished product. It is one of humanity’s oldest and most important forms of communication. When people can’t fully express themselves through words, they often find a substitute through drawings, music, or movement. Children communicate emotions through art long before they have the vocabulary to describe them. Art therapy helps people process grief, anxiety, and trauma by creating rather than explaining. Human-made art always carries traces of the person who created it.

Most people recognize the value of human-made art. Participants in a study conducted by Columbia University’s Business School showed significant bias against AI-generated art, rating it lower on markers such as creativity and the amount of work involved. At the same time, we can quantify just how bad humans are at detecting AI-generated art from human-made art. This study, published by the Association for Computing Machinery, found that people correctly identified AI-generated images only about half the time, which makes it no better than randomly guessing.

Despite hesitancy and bias, AI-generated art has undergone a rapid expansion in recent years. In 2022, an AI-generated image won first place in the Colorado State Fair digital art competition, sparking widespread controversy because other competitors had created their work entirely by hand. And in 2024, Ai-Da, a humanoid robot, sold an AI-generated portrait of Alan Turing for over one million dollars. AI-generated art is driving the growth of the art market globally, selling at top-tier auctions, and winning prestigious competitions. In some of those instances, AI art is clearly labeled as such; in others, it isn’t. But if people cannot tell the difference, they cannot make informed choices about the art they consume. 

Clear watermarks on AI-generated images would solve this problem by making the source of an image immediately visible. What’s more, widespread watermarks would protect artists. While it’s true that thousands of creators now use generative AI in some part of their work, artists who haven’t sometimes face online harassment after being falsely accused of using AI, all because their work has tiny mistakes or is designed in an “AI art style.” Their reputations suffer despite them having created every stroke themselves. A clear watermark indicating AI-generated images would reduce these false accusations by giving viewers a reliable way to distinguish between AI-assisted and human-created work. Watermarks would also help maintain fairness in competitions: if AI-generated works were consistently and accurately labeled, competitions could establish and enforce clear rules about eligibility instead of relying on assumptions or good faith.

 My AI-generated hummingbird was beautiful, sure, but it never reflected anyone’s thoughts, memories, or experiences. The sunflower I drew for my mom was different; every choice I made came from the way I see our relationship. Viewers should be able to tell whether a work of art was made by a human artist or produced by AI. Clear watermarks won’t diminish AI’s potential. Instead, they will preserve our ability to recognize and appreciate the human stories that art has always been meant to tell. 

Jane Zhang is a 9th grader at Buckingham Browne and Nichols School in Cambridge, Massachusetts.