DALL-E 2: The Rise of AI Painting Generators
Words by Edsard Driessen
One of the most shocking discoveries I have recently stumbled across while mindlessly scrolling through Instagram and TikTok is the concept of AI-Generated artwork. Often introduced with an eye-catching description such as “Donald Trump in Washing Machine” or “The city of Atlantis in Watercolour”, subsequently then painted by an AI programme in mesmerising detail and vivid colours. The scope and opportunity of these online, often free, programmes is boundless and the results and consequences for the world of art and painting are even more so.
AI Generated artwork of The Lost City of Atlantis
Given the almost instantaneous beauty of some of the cleverer creations offered by users of these programmes, the prospect of a digital canvas painted with no user skill or creativity required certainly presents a new dawn of artistic opportunity. However, as someone who appreciates art, the prospect of robots taking over the creative process worries me and leaves me wondering what this means for the world of human creatives.
Using an AI tool like DALL-E, all you need to create an image is to describe it in text and the programme will produce an image that matches your description. These AI text-to-image generators are powered by hugely complicated algorithms that can be trained and ordered to produce the desired paintings. On the surface level, these generators seem to spell the end of the creative world as we know it; how can an artist be revered for a piece of art when anyone can create something just like it with a few clicks of the mouse? However, Anne Ploin, from the Oxford Internet Institute claims that AI art generators won’t replace human artists; she argues that rather than eclipse human artists, they will be used merely to assist creatives in speeding up the production of high-quality work. If this doesn’t sound particularly convincing, Ploin further makes the point that a total eclipse won’t happen any time soon as the creative decision-making aspect of artistic production cannot be replicated by current AI technology. In this case, Ploin points towards the inability of AI to be able to reproduce creative thinking, given that the generator programmes are still reliant on text input to create works of art. Given the wide consideration needed before any professional artist puts brush to canvas, Ploin argues that no artificial intelligence system could ever reach such conclusions.
AI generated MTG artwork
Not only in the creative planning for a painting does an AI programme retain its position as a mere “assistant” to the creative process, but also in the personality and outlook of an artist does it fall short. Especially nowadays, an artist showcases themselves as a personality, where their work comes to be a mirror of that fact. An AI generator showcases only an artwork or painting and could never demonstrate any concept of expression, sentiment or individuality. The AI programme therefore acts within a very one-dimensional idea, where they can only re-create, or replicate images put together through a series of datasets collated through its algorithm.
Through the use of AI-controlled art generators the artwork loses its human emotion and individual connection, but this does not mean that it is not art. Having highlighted a few reasons above as to why the AI generator will never surpass the human artist, I must suggest the idea that this may not be entirely the case. These programmes still produce art: art which is even winning prizes over human competitors. This is highlighted in The NY Times, which reported Jason Allen’s win at the Colorado State Fair for his AI generated art work titled “Theatre D’Opera Spatial”, winning over hundreds of other entrants, all of whom did not use an AI generator. Using Midjourney, an intermediate AI generator programme, Allen created a digital artwork without the use of a single brushstroke. Understandably a win like this speaks volumes about the current state of the creative world; an AI programme may never be able to replicate individuality or human emotion, but it certainly can produce artwork capable of winning over its human competitors. With this in mind, we must then consider what is most important?
Theatre D’Opera Spatial by Jason Allen
Given the early stage of these programmes in general, it is hard to draw a finite conclusion on the matter. However, we ought to be aware of this new development considering that the creative world may change for good with the consistent use of programmes such as Midjourney or DALL-E.
