ArtMeta

Intelligence

Machine Learning, Agency and the Question of Authorship

Back to Catalogue

From Program to Model

With the transition into the 2010s, a profound shift began within digital art: in place of code explicitly written by the artist came the trained model. The new image systems are based on neural networks that learn patterns from large quantities of data and generate new images independently from them. The role of the artist changed fundamentally: no longer formulating instructions alone, but working with datasets, training processes, model architectures, and prompts. Authorship shifts from direct form-giving to the design of the conditions under which machines produce images.

This development depended on new technological infrastructures - advances in deep learning, cloud-based computing resources, and the GPU architectures driven by companies such as NVIDIA. What universities and industrial research laboratories were in the early days of computer art, global technology companies and AI research groups now assumed. Yet alongside this infrastructural shift, a parallel development took shape: a series of artist residency programs and technology-embedded laboratories that deliberately created space for artistic experimentation within, or in close proximity to the technology industry itself.

The LACMA Art + Technology Lab, relaunched in 2014 as a conscious echo of its original 1967–1971 program, which had brought artists including Robert Rauschenberg, James Turrell, and Roger Hiorns into contact with aerospace and computing industries, invited artists to engage with emerging technologies such as augmented reality, artificial intelligence, and autonomous systems. Similar structures emerged at Google with its Artists + Machine Intelligence program (AMI, founded 2015), which defined machine learning explicitly as a creative medium and in collaboration with Google Arts and Culture, supported artists, researchers and cultural practitioners working at the intersection of art and artificial intelligence. NVIDIA, whose GPU architectures made contemporary deep learning practically viable, became a visible cultural actor through its AI Art Gallery - a curated platform spanning generative AI, GANs, NeRFs, and immersive systems, in which artists and developers are presented alongside one another.