Can computers be creative? Is algorithmic art just a form of CandyCrush? Cutting through the smoke and mirrors surrounding computation,robotics and artificial intelligence, Joanna Zylinska argues that, tounderstand the promise of AI for the creative fields, we must notconfine ourselves solely to the realm of aesthetics. Instead, we need toaddress the role and position of the human in the current technicalsetup – including the associated issues of labour, robotisation and,last but not least, extinction. Offering a critique of thesocio-political underpinnings of AI, AI Art: Machine Visions and Warped Dreams raises poignant questions about the conditions of art making and creativity today.
91 chapter 8 algorithms. This interrogation is important because, as shown by safiya noble in Algorithms of Oppression: How Search Engines Reinforce Racism, data and computing have become so profoundly their own truth that even in the face of evidence, the public still struggles to hold tech companies accountable for the products and errors of their ways. These errors increasingly lead to racial and gender profiling, misrepresentation, and even economic redlining (noble 2018). critical projects such as those by paglen encourage us to ask: Whose vision is ai promoting? Who is doing the looking, in what way and to what purpose?15
id: a404cd19b072e8473790879667248ba4 - page: 93
a collaboration with Kronos Quartet and light installation company obscura digital, it involved staging a concert in a san Francisco warehouse, accompanied by the projections of various bits of data driven by ai algorithms. as well as displaying, in frequent motion, one of the face recognition training data sets discussed above, the artist had installed a number of cameras in the warehouse, with feeds going into the video mixer and the hardware. The cameras then made visible on the screen behind the band renderings of outlines of the human members of the band in the form of multicoloured squiggles, circles and squares. The artist and his team occasionally turned the camera on the audience to allow them to see themselves being seen by the computers, with their faces identified as faces and also rendered as rectangles. The idea behind the project was to examine the 92
id: 8258c098dbcb5ee870e0e648a3bf7b11 - page: 94
Yet we should ask to what extent this is actually still seeing. and are we still talking about intelligence? or are they just behaviours that look like seeing and intelligence to us, their human interpreters?
id: ac9eb0eef1ee729b4807b91032d42e65 - page: 95
, these questions are important because the systems put in motion that enable computer vision and other forms of ai determine who and what is allowed in and what isnt. Much translation work had been involved in paglens Sight Machine, but at the end of the day the performance revealed the basic untranslatability of data between different recipients, resulting from the opacity of code (from brute force algorithms of the 1960s systems to contemporary deep learning frameworks such as TensorFlow, Torch and caffe). it is precisely in that very gesture of attempting to undertake the work of translation that the incompatibility between different cognitive frameworks and different forms in which intelligence is embodied was revealed. The project thus succeeded and failed at the same time: it failed at transparency, at revealing (to us) what and how computers supposedly see, but it succeeded at unveiling this translat
id: d43e76a6463d0cc6c99469ca6368f911 - page: 95