How to Read an AI Image

Organized by NYU Digital Theory Lab

EVENT Photographs by:

TANNON Reckling & Marina Hassapopoulou

EVENT DESCRIPTION: How To Read an AI Image: Workshop on Diffusion Models with Eryk Salvaggio 

Friday, October 20th, 2023 – Hybrid

Artist & researcher Eryk Salvaggio will present an overview of how Diffusion models are trained and then produce images. This will serve as a framework for thinking through entanglements of social and technical bias in these systems. Finally, Salvaggio will walk through a methodology that uses this framework to understand and interpret AI images as “infographics about the dataset.” In the workshop portion, guests are invited to share prompts for images to apply this method against, practicing a method of analysis and interpretation that hopea to illuminate how meaning is made in these images, and how central tendencies of datasets manifest themselves visually. 

Our impressions (more to be added soon): A thought-provoking talk by researcher-designer-artist Eryk Salvaggio, accompanied by a hands-on exploratory workshop led by Eryk on the limits and ethics of generative AI. The post-talk workshop provided the opportunity to put theory into practice by engaging in various activities that tested the limits of the AI imaginary. Collectively testing some of the “micro” aspects and particularities of generative AI and Diffusion models helps us grapple with some of the “macro” aspects of AI and to carve out room for human and ethical intervention. The shortcomings of generative AI correspond to larger issues regarding cultural and systemic bias… More thoughts coming soon…

You can read more on reading AI images on Eryk’s website.

Some photos from the event:

Leave a comment