By Leonardo Mincolelli – Florentine Photographer. Cover image: Naufal Farras.
AI and photography have often generate confusion and misunderstanding. While artificial intelligence has quickly evolved and integrated into our lives in ways we never expected, its role in creative fields is still a topic of debate. For years, AI was viewed as a distant concept in academic discussions or portrayed in the media as either a technological marvel or an existential threat. These contrasting perspectives have sparked a mix of fascination and fear, causing people to either embrace or reject AI in their everyday lives.
As a cognitive science student, photographer, and artist, my goal is to address these concerns and offer a calmer perspective, arguing that artificial intelligence should neither be feared nor overly revered but rather seen as a tool to be used with intention.
AI and Photography, Image Generation Through Diffusion Models
In this discussion, we will focus on a specific AI model commonly used for image generation within the fields of AI and photography: diffusion models. Essentially, this term refers to models trained on the principle of noise diffusion — similar to the Gaussian noise effect in Photoshop—applied to an existing image that has been retrieved from the Internet based on a text search. This image serves as the “prompt” guiding the AI’s interpretation.
When we ask the AI to generate an image of a horse, it draws from a vast dataset of images it has collected from the web. The AI analyses and deconstructs these images to extract the visual data necessary for creating a new image. By adding more descriptive elements to our prompt—say, asking for an astronaut riding a horse on the moon—the AI performs a type of collage. It identifies the horse as the only common element in all horse-related images, and then it places it against a lunar backdrop, with the astronaut on top. In doing so, it mimics known representations of human horseback riding and astronauts.
AI in Photography and in Public Debate
AI includes a variety of models and subcategories, each specialising in different areas like text generation, audio production, or data processing. Its impact on the job market, particularly in media production, is already evident. At this point, most of us have likely come across content that was either partially or entirely generated by AI.
On one side of the debate, AI advocates—especially those who benefit financially from it—highlight its accessibility and how it empowers millions of users to create content without needing extensive tools or specialised skills. In contrast, professionals who have invested time, effort, and resources to acquire those skills view AI as a threat that could make their expertise obsolete.
This is the mainstream narrative that dominates media discussions. However, I believe there are deeper implications that are often overlooked—insights that become clear when we analyse the process by which AI “generates” its outputs, as mentioned earlier.
AI Transforms, Humans Create
First and foremost, no AI model can create something from nothing. Every existing and future model, including those employed in AI and photography, will always be constrained by a fundamental lack of understanding regarding what makes something aesthetically appealing to humans. AI not only requires extensive training but also lacks the foundational principles of aesthetics—concepts that have been explored and refined over centuries by countless scholars. More critically, these principles are not yet fully understood by humans themselves.
This limitation makes it not just difficult but indeed impossible—given our current knowledge of cognitive science and the philosophy of aesthetics—to design a machine capable of independently grafting something that is both intelligible and emotionally resonant without first analysing human-made works. AI does not comprehend composition, meaning, or beauty in the way that humans do; it simply identifies patterns based on the likelihood of certain colours, shapes, and arrangements appearing together.
While it’s true that humans also need to hone their creative skills—often by learning from the works of others—the key difference lies in how and why we learn. Some argue that AI evolves similarly to human civilisation, accelerating cultural evolution. However, research shows that AI quickly hits its limits, which are typically defined by existing human creations, undermining the belief that it can foster innovation in the same way that people do.
AI Is a Tool, Not a Replacement
In conclusion, while AI can deconstruct and recreate human-made works—an essential function in the commercial side of creative industries—it can never truly replace those who create not just by drawing on what exists but in spite of it, in the pursuit of genuine novelty and innovation.
It is crucial that this message reaches the next generation of creatives, who may feel discouraged by the perceived threat of AI and consider giving up on their ambitions—an outcome that would be a great loss for humanity as a whole.
