What Defines a Photograph?

Photography involves two essential elements: a technical aspect and a conceptual one. The first element is light. At its core, a photograph captures a specific moment of light that is preserved indefinitely. The term "photography" literally means "drawing with light," which requires a profound sensitivity to the nuances of light.

Light can evoke emotional responses, often subconsciously. For instance, a portrait taken during the golden hour radiates warmth and evokes different feelings compared to one shot under the harsh fluorescent light of an office.

Throughout the history of photography, methods have evolved—from the use of polished metal plates to modern digital sensors—yet the essence remains unchanged. Photography is fundamentally about capturing and preserving real-world light.

The Concept of Pointing

The second critical aspect of photography is conceptual. As curator John Szarkowski articulates, photography is "the act of pointing." This definition resonates deeply with me. In my photography classes, I encourage students to consider the message behind their images. If the answer is merely "Here is something," then the photograph lacks depth.

For example, in food photography, an uninspiring shot may simply present a dish without context or emotion. In contrast, a compelling food photograph might emphasize a key ingredient, the ambiance of the dining environment, or the feelings associated with shared meals.

Ultimately, the artistry of photography lies in the photographer's ability to highlight and convey meaning from the chaos of the real world.

AI Images Fall Short

AI-generated images, however, miss both of these critical elements. Firstly, they do not employ light in a genuine sense; rather, the light in these images is simulated, created by algorithms in a distant server. The creators of these images often have limited control over the lighting, as AI tools like DALL-E or Midjourney determine the lighting based on generated scenes.

For instance, if you request an image of a beautiful person, you are likely to receive an illustration bathed in the soft light of dawn. Conversely, a request for a hacker may yield a figure shrouded in dim, red light.

Moreover, the simulated light in AI images is influenced by the training data, which often consists of vibrant, eye-catching images from social media platforms. One notable project, "Boring America Photorealism" by John Kelly, attempts to use AI to generate mundane images we often capture on our phones, illustrating the challenges involved in achieving realistic, unremarkable results.

With traditional photography, the outcome is determined by real-world lighting conditions. In contrast, AI images are created by designing a subject first and then constructing a surrounding context based on what seems appropriate, rather than highlighting a real-world object or concept.