A Selection of Recent Workshops: 

Workshop took place on Ikon Slow boat with Ikon Youth Program, Birmingham, UK, August 2024, Photo credit: Cathy Wade

​​​

​

One approach to working with and understanding AI as an artist / educator is to intentionally introduce errors into the system. This methodology references artists such as Eryk Salvaggio who uses AI tools to generate digital noise - overriding AI designers' intentions to create neat infographics of data training sets. 

 

This assignment invites participants to creatively misuse Artificial Intelligence tools by breaking up the form of the body using collage materials. It also gives participants an opportunity to question algorithmic cultural biases. By intentionally introducing errors into a system we can begin to explore and exploit glitches in the classroom. These approaches are sometimes called Adversarial Attacks, or Data Poisoning. Many facial recognition systems use neural network type approaches to detect patterns in large databases of images of faces. The large field in an image is broken down into quadrants, this detects and puts a bounding box which is then run through a deep neural network and through other measurements such as classifications, clustering and against identification set in order to create facial recognition. Apart from privacy issues, this system can create errors and there are unintended consequences to mistaken identity. One anecdotal example here is of a business woman in China being fined and publicly shamed for J-walking when in fact her face was used in an advertisement on the side of a bus (Rose, 2019). These surveillance systems can also be intentionally fooled or tricked by wearing simple disguises. 

 

Methodology:

 

In pairs cut-out generated facial features and metallic vinyl and attach these to your face. Take it in turns to draw one another. Use the drawing as a prompt to create a new generative image. Print the result and restart the process but this time using the recursively generated image.

​

​