Training Set (currently in progress)
'Training Set' is a new project that uses absurd humour to engage people in current debates around the use of Artificial Intelligence as a solution to difficult and complicated human problems such as loneliness and anxiety. It is envisioned as a sculptural and audio installation featuring silicone casts of women’s mouths that conceal microphones and speakers connected to the internet. Visitors can talk to these disembodied mouths - they use Voice User Interface (VUI) technology, the system used by Apple’s Siri or Amazon’s Alexa that allows people to ask for information or give instructions using their voice.
‘Training Set’ mock-up sketch
Each mouth sculpture will be ‘trained’ to offer conversations on specific themes and topics (known in VUI terms as ‘custom skills'). They allow the person to drive the conversation by automatically retrieving information from the internet in real time. Each interaction is unique: the VUI learns and builds on conversations with each visitor. Conversation topics in ‘Training Set’ will be based on concepts such as emotional labour (feminised 'soft skills'), reproductive health and mental wellbeing. For example, one talking mouth may be able to offer you the latest advice on freezing your eggs or on where to find a sperm donor, while the next offers a complementary therapy session.
‘Training Set’ mock-up sketch and test cast (alginate, pigment)
The project attempts to call into question the way in which this new technology reinforces harmful gender stereotypes. From the first VUI program ‘ELIZA’ in the 60s, to today’s Siri and Alexa, voice assistants are feminised both as a tool of soft persuasion and due to unconscious bias - implying that female means somehow less than equal (as subservient assistant or as a machine/child/pet that requires training). It also asks the question of who is training who. VUI technologies are becoming unavoidable: they are integrated into more and more public services to replace a paid workforce, and many people have to adapt their natural way of speaking to be understood by the machine. But on a deeper level, how are algorithms training us via attempts to optimise our capacity or willingness to produce and consume in the techno-capitalist society we find ourselves in?