What is AI?
Artificial intelligence (AI) is the simulation of human intelligence processes by machines, especially computer systems.
Image Captioning and AI: A Brief History
“Image captioning” was a major development in AI research that occurred in 2015 which, in broad terms, means labelling images and adding a description to them.
For the sake of curiosity, the mechanism of “image captioning” was flipped around to come up with “text to image”.
The idea was to create novel scenes, outside of the real world. Initially, AI’s depiction of art from the texts with unusual descriptions were mimed through 32 pixels of blurry, abstract images.
Altogether, creating a certain type of art required a collection of a specific dataset of images and training the model to mimic that data. Therefore, just as humans have various techniques of their own to create any form of art, different AI models give different results based on different data sets. It depicts that various models have their individuality as well, because each of the software has been designed by different people, i.e., AI itself is a work of human creation.
Human Art and AI Art: Key Differences
According to Plato’s viewpoint, “poets are twice removed from the world”; in a more realistic situation of the current scenario, Artificial Intelligence dramatically separates us from the ideal, and existing world in the present time, as it is specifically trained and developed to ‘mimic’ what has been present all along. On the other hand, art by humans might take us back to that ideal world but through an out worldly journey of their imagination, as human artists’ purpose is to create something different from the ordinary, to add originality. This makes both the art types incomparable.
Latent Space
Another aspect of AI generated art images is ‘latent space’. “It is a multidimensional vector space that represents the learned characteristics of a set of data. This abstract space encapsulates the essential features and attributes of the input data, fostering the development of efficient AI models.”-- Lark suite.
The latent space of these AI models contain some dark corners including violence. There are multiple sources on the internet that contain misinformation or include gore substances which help shape the datasets for an AI model. These factors get scarier and avoidant as the outputs become photorealistic.
Humans are incapable of imagining a multi dimensional space, but midjourney’s model of generative AI offers the same and makes it possible to study and collect multiple variables to produce a certain image.
This can lead to copyright issues as the original artists of the “data” that helps the formation of AI models become sceptical as their jobs are threatened and work is stolen to develop another model.
Art and Emotions
A noticeable point is that any work of art by humans is done by hand. It consumes time, effort, and thought. The artists have an emotional connection with their work. On the contrary, AI is a software machine, and AI generated art does not involve the emotional factor and therefore, has less meaning to it.
Conclusively, AI generated art is still art as it has been trained through multiple datasets but what separates it from the art of any sort by humans is the lack of emotions put into it, the experiences and the existence of humane presence in itself.
My final note is that AI generated art can never be compared to the art made or formed by humans and therefore, it is not “real art”.
-Jiya Kumari
BA (Hons) English, Bharati College, University of Delhi
Jiya is a peace loving and introverted individual. With a natural inclination towards solitude, she finds solace in the pages of a good book, a coming of age film, and the subtle art of people watching. Despite a preference for quiet reflection, her curiosity leads her to exploring new hobbies and being committed to personal growth.
Comments