The Big Smoke

2023

Machine Learning

Béatrice Lartigue - Lab212

Wednesday 7 June 2023. New York wakes up to a thick orange fog covering the city. It's the forest fires in Quebec that are darkening the skies over the Big Apple, 800 km to the south. The intensity and frequency of these fires are linked to climate change. The concentration of micro-particles is ten times higher than the standards recommended by the World Health Organisation.

Today. The IPCC reports that "Increases in the frequency and intensity of several types of extreme weather and climate events (heat waves, heavy precipitation, droughts, etc.) are having irreversible impacts by pushing natural and human systems beyond their adaptation limits in all regions of the world". 1

1 Climate: synthesis of the 6th IPCC report, The Shift Project

How can we use AI to explore our relationship with our environment in a speculative and forward-looking way?

The expression "The Big Smoke" evokes a big city and the pollution associated with it. The Big Smoke takes visitors on a journey through time and space, with a central theme: air pollution.

The project draws on Design Fiction through a speculative and critical approach aimed at inspiring alternative visions of the future, by materialising potential scenarios. The deployment of generative AI opens up an imaginary world around climate change and encourages the exploration of related concepts (urbanisation, density, flows, etc.).

The Big Smoke combines research and simulations in order to materialise non-visible elements, explore scenarios, encourage visitors' curiosity and provoke discussion. The simulations images are generated with text to image machine learning model.

The aim of the project is twofold,

• raise awareness: to provide a tangible way of visualising intangible data (pollutant levels detected by measuring stations, patterns: levels change according to daily activity, season, GPS position, etc.).
• take the drama out of it: to question our collective imagination and open up the discussion of possible alternatives (soft modes of transport, greening, insulation, etc.). —

The artistic vision is deliberately joyful and singular. The gravitas of the subject is faced with this paradoxical representation, creating a shift that enhances the issue and prevents rejection at first sight.

The Big Smoke invests the codes characteristic of architectural photography (perspective, textures, depth, etc.) and contemporary photography. The city presented is a fictional vision of the city of Paris, with similar characteristics (Haussmannian inspiration, Art Nouveau decor, post-industrial architecture, etc.), but which has no existence of its own.

The embodiment of atmospheric pollution in a thick coloured mist plays on the metaphor of the cloud, while at the same time enhancing the dramaturgy of the images (contrast between the urban context and the coloured atmospheric phenomenon). The artwork draws on the colour code of the forecast maps to modulate the tint of the smoke clouds (cold/good > hot/extremely bad). The project mines air quality data, through an overall index including ozone, nitrogen dioxide, particulate matter and fine particles.

Finally, the light processing specific to today's generative AI algorithms lends a realistic and authentic dimension to the corpus of images.