Artificial Intelligence as an Artistic Tool
OpenAI
Gaz Lawrence has been using AI in some of his work since early 2022. Initially part of OpenAI’s Artist Assistance program. In exchange for feedback, Gaz has had free and early access to some of their cutting edge AI technology such as their latest text-to-image generator DALL-E v2.Exp which you can learn about in The MoSAIC virtual museum. Lawrence believes that generative AI is a vital tool in the artist’s toolkit in the post-AI art world.
Skipping
Ink on Paper
2021
Skipping AI Variations
Spray Paint on Canvas
2022
Originally inspired by Francisco de Zurbarán’s paintings of monks, ‘Skipping’ started off life as an ink drawing used as the single artwork for the electronic musician Absconda’s track of the same title. The image then went onto become a painting and most recently the image was manipulated using DALL-E 2 text-to-image AI, trained by reinforcement learning to draw in Lawrence’s style. This most recent version makes up The MoSAIC x Gaz Lawrence collection. The collection highlights how AI can help to facilitate human creativity, rather than usurping it. DALL-E was initially used to create a variation on the original ink drawing of ‘Skipping’. Lawrence then edited this image on Photoshop before running it through the AI again. This process was repeated ten times before the image below was created. Many people have reported seeing double images in the resulting artwork. What do you see?
For the final artworks the AI generated image has been reproduced with the use of either block printing or stencilling in a range of editions
Exactly.ai
The above images were produced by a different method. Perhaps one that will prove more useful to Lawrence’s future work. The program Exactly.ai (originally called Durer.ai) can be used to train your own AI text-to-image model. By inputting between 5 and 15 images of your own artwork and describing them in natural language you can train Exactly to create works in your style. On the left, we have an output from the Durer x Gaz Lawrence text-to-image model. This allows Lawrence to digitally play with different themes and compositions before producing the final artwork. On the right is the final ink drawing which took over 20 hours to complete with just a dip pen and India ink. AI is incredibly useful in augmenting human creativity. Artists can explore the limitations of AI in order to produce artwork that is uniquely human.
The above images were generated using the Exactly.ai x Gaz Lawrence model and were exhibited alongside other artists using Exactly at the Shoreditch Arts Club, London on Friday 8th September 2023.
You can now try generating images with this AI model for yourself.
Ghosted II
Oil and acrylic on canvas
60cm x 60cm
Based on a study from 2021, Ghosted II is an exploration of the unique feeling of being ghosted. An abrupt end without satisfactory explanation. This artwork delves into the emotional dichotomy of intimately knowing someone and then facing their sudden absence.
The painting captures the essence of this transition, from connection to loss. The complex doodle pattern behind the anonymous figure is a distant memory of a complex relationship and can represent either party in the divide. With their back to the viewer, they now look forward into a void that may look saturated in colour at first but is drained of the intricacy of what has been left behind.
This painting is an experiment with the Exactly.ai x Gaz Lawrence 2023 text-to-image model. The model was trained on Gaz’s artwork, including the 2021 study for this painting. Using natural language, the final rückenfigur composition was developed and then reinterpreted on canvas by hand using oil and acrylic paint to draw the viewer into the feeling of isolation and longing.