Unaccountable AI

Joseph Anthony Connor

United Kingdom
Artificial Intelligence Award for Digital Art, sponsored by BCS, The Chartered Institute for IT
Unaccountable AI Emergency Triage language

Artificial Intelligence ‘Black Box’ (AI) algorithms defy easy categorisation.  They can  constantly innovate from the data humans leak from text, phone calls and phone usage.  In 2017, there has been much talk about the subject of AI, but little communication about the impact of its use.  This artwork aims to do that.  It is based upon emergency triage dialogue in 2016.  In the work’s first white box, the more frequently a human leaked the word the larger the text.  In the second white box, the more frequently a word is used in a Black Box algorithm the larger the text.  As the viewer will see, the link between the two is not apparent.  More importantly, no human could explain the link.  It is therefore unaccountable.  This is worth bearing in mind when you think about what AI means to you.

NVIDIA CUDA, Tensorflow, Cloudart and Phoetic