Contextualising AI: Sampling the Masters

13, 16 December 2022

Course Brief

Think about the masters of craftsmanship you resonate with in different timelines throughout art history. Using the text & image-based AI models, experiment with your own narrative exploring the idea of artists’ ownership.

Copy, modify, break down, re-generate, re-purpose, mix, stretch, zoom in and out, collage, tear it apart and re-stitch the original artwork using AI models. Question yourself what this means to you.

This workshop is designed to critically about the topic of artificial intelligence and creativity by challenging the concept of authorship.

After understanding the methodology of Sampling and discussions about ethics in AI, we touched upon and experimented with tools such as GPT-3, dall-e, Hugging Face Transformer, Disco Diffusion, Stable Diffusion, Flow Frames, Melobytes, and Ebsynth.

After a long day of learning, the students gave presentations about what they researched and what direction they were interested in developing.

Students’ interests were fascinatingly varied, from Tunisian rap images (Ines), flies, metamorphosis, failures of communication (Bart), and K-pop stars (Seungun). A student, Werd, collected pictures from SD cards he found in secondhand shops to make a dataset to train.

After experimenting and producing, we had a small laptop exhibition of all the works.

Sophie created the ‘cursed’ version of Spongebob.

Lukaz reworked Keith Harring.

Sangeun created an imaginary idol’s album cover to critique the artificiality of the K-pop industry.

This workshop went over time with enthusiastic students discussing the works. We went for a beer afterwards to celebrate! Thanks to all students being tightly engaged with learning!

More workshops & talks?

  • Contextualising AI: Sampling the Masters

    Contextualising AI: Sampling the Masters

  • Augmented Projection with TouchDesigner

    Augmented Projection with TouchDesigner

  • Contextualising AI: The Blue Fairies

    Contextualising AI: The Blue Fairies

  • RGBdog presents: Volumetric Interviews

    RGBdog presents: Volumetric Interviews