Artificial intelligence is a technology that has stirred the art industry. Whilst on work experience, Shael Kumar (Yr 11 student) explored the implications of code meeting canvas, as AI continues to redefine artistry.
Art has been democratised. An art piece that would have taken a professional hours to complete, can now be made in a few minutes by anyone who has an internet connection. This astounding feat has been made possible due to one pivotal technology: artificial intelligence.
The Development of Generative Art:
Recently, AI generated art has flown into public consciousness, from out of the blue. The excitement surrounding this technology has grown exponentially, with services such as DALL-E and Midjourney showcasing its great potential. However, these ground-breaking AI models have had some humble beginnings.
In 2015, early AI models such as AlignDRAW were built using simple neural networks that outputted basic visual patterns and rudimentary compositions. This meant they were limited in capability, acting as minimal supportive tools for artists. Nevertheless, these models laid the foundation for subsequent advancements in generative art and were proof-of-concepts that AI models could emulate the artistic capabilities of humans.
These developments caught the eye of Sam Altman (the CEO of OpenAI), among others. With investment from influential personalities such as Elon Musk, OpenAI facilitated advancements in machine learning and deep neural networks, using substantial amounts of data from the internet. This has led to breakthrough innovation, augmenting technologies such as stable diffusion, that have propelled AI art into the realm of detailed and autonomous creation.
Similarly, generative adversarial networks (GANs) have further transformed AI art by allowing for the creation of realistic and visually appealing artworks. GANs are composed of two neural networks: a generator that produces images, and a discriminator that evaluates their quality. Due to this, AI art has become more alluring and captivating in the eyes of its beholders. In fact, the art created by AI is so incredibly aesthetic, that it has won various competitions, such as the Colorado State Fair’s annual art and the World Photography Organization's Sony World Photography Awards.
Ultimately, generative art has had a rich developmental history. The incredible results we are seeing at present are the fruits of software programmers’ meticulous labour.
In the past, new technologies have generally allowed for greater creative freedom in art. When applications such as Adobe Photoshop were released, they initially received some resistance from traditional artists, as there were concerns about the effect these apps would have on the authenticity of artistic expression. Eventually, however, these applications have been adopted by the main-stream art community, and have led to enhanced productivity.
With the advent of generative AI, a similar situation is unfolding. Generative AI allows for the symbiotic fusion of human creativity and computer algorithms. Far from replacing artists, AI helps them feel more fulfilled. This is because it takes away the tedious aspects of artistry, and instead gives artists more time to invest into genuine creativity. Therefore, AI allow artists to be more efficient and productive than ever before.
Moreover, the artistic liberation that AI provides is unparalleled. The technology opens the door to artistic expression, for practically anyone and everyone. AI makes artistry incredibly accessible and simple. It so highly intuitive, that even if one has no background in art, they can instantly materialise their thoughts into beautiful art for free, using sites such as Craiyon.
Generative AI also offers advantages to businesses and brands. For example, Midjourney is actively being used by entrepreneurs for product shots, as it significantly reduces the complexity of the process. AI algorithms can generate stunning visual representations of potential products, simply based on the prompts inputted to them. Not only is this great for rapid prototyping, but also provides a great medium for the personalisation of mood boards. Moreover, AI will be very useful who business who market themselves using YouTube since generative art can be used for the swift creation of thumbnails and even background, stock footage. Overall, AI art will lead to a significant productivity boost for all types of businesses.
Hence, AI generated art has created some exciting new opportunities that if utilised properly, can be used to society’s advantage.
For any new technological advancement, a balanced understanding is imperative. This is especially true for artificial intelligence, because if it goes in the wrong direction, the consequences could be grave for all of humanity.
Firstly, in the short-term, the training for these generative AI models has been done using images from all over the internet. While companies such as OpenAI have made sure to exclude any inappropriate or politically inflammatory images, they have trained their models on copyrighted artworks that were used without the consent of the artist that originally created them. Some artists have even gone to court to sue these AI companies because of this. Nevertheless, these AI giants may argue that their actions are protected by the “fair use act”. Due to there being a lack of a precedent at the moment, the pendulum can swing in any direction.
In the long-term, we could face an even more existential threat. The AI art generator DALL-E was built using a subset of the GPT-3 LLM (Generative Pre-trained Transformer 3 Large Language Model). As this underlying GPT improves, we could reach a technological singularity; a digital superintelligence that is far smarter than all of humanity combined, could be created. When this occurs, it could either be a blessing or a curse. To make sure that the former outcome is the most likely, we must develop AI in a prudent manner, in line with recommendations from relevant authorities across the world.
Consequently, whilst AI art and its foundation technology are undoubtedly fascinating, we need to be considerate and careful as we develop it.
Accordingly, with the increased investment into AI, generated art is likely to get even better. Google has been investing heavily into DeepMind, while OpenAI has continued to release new versions of GPT. The newest model, GPT4, has more than 170 trillion parameters (the number of weights in the connections of a virtual neural network), which is akin to the number of neuron connections in a human brain!
Indeed, as Károly Zsolnai-Fehér, the man behind the “Two Minute Papers” YouTube channel saliently said, “what a time to be alive!”
Year 11 Student
Bishop’s Stortford College
6th of July