Public Parts is situated within the context of digital platforms and the precariousness of life associated with the rising informality of work. In the past 15 years, we have witnessed a proliferation of platforms that deal with new non-specialised occupations that constitute the gig-economy. Gig workers often lead a precarious life of underpaid, unrelenting tasks. Paired with the urban housing crisis, this demographic is left unsupported and often rent-burdened.
What if digital platforms and task-based work could be used to empower the autonomy of these communities?
Public Parts is an AI-managed platform for autonomous communal living. It seeks to subvert neoliberal platforms that currently control the gig economy, with a re-socialised stance on social housing. The platform utilises automation and machine intelligence to produce buildings capable of spatial reconfiguration to be inhabited and managed by housing cooperatives.
We seek to discretise and decentralise the typical apartment in order to minimise fixed ownership and maximise access.
Public Parts’ AI manages a gig-based construction of discrete parts, automated spatial reconfiguration, communal upkeep, and a domestic and professional task pool which operates as a socio-economic infrastructure.
The result is an environment that adapts to the behaviour of its inhabiting community while providing gig work opportunities.
Through the resocialisation of social housing, gig-based work, and automation, Public Parts proposes an AI-managed platform for autonomous communal life.
The platform operates as a circular economy, with an open-call task pool consisting of upkeep, domestic, and professional tasks. These tasks are picked up by individuals in exchange for money.
Imagining resocialised life raises crucial domestic questions of what is owned, how and to what degree privacy is achieved, and what remains fixed in a configurable platform for communal life.
Spaces can be recofigured through both tenant feedback and machine learning.
Pix2Pix is utilised to differentiate spaces in a Public Parts building. Pix2Pix is a conditional generative adversarial network that allows the AI system to recognise spaces through user biases.
Configurable parts are defined as machine learning agents. Trained through reinforcement learning, they collect observations and follow discrete actions to reconfigure spaces.
Virtual tenants and configurable parts allow for the evaluation of the designed behaviour of the building’s reconfiguration. A four story Public Parts building serves as proof of concept.
The distribution of these parts relies on a simple rule-based algorithmic logic which aggregates parts in particular ways to create space and pixelated ornamental structures.
Further research within the rule-based organisation of parts led to a prototypical building for Public Parts. This building is based on a central megalithic core and distributed living rooms.
The platform’s preliminary case-study architecture operates at the scale of a multi-family villa which could host a small population of tenants.
A porous bar-block typology containing community gardens, platform programmes, and reconfigurable space, made entirely of discrete timber elements.
Through automation and architecture, Public Parts establishes a socio-spatial scenario in which individuals produce the culture of autonomous communal life, labour, and leisure.