MRI of the Earth by Refik Anadol

Through 3 chapters, MRI of the Earth explores extreme climate-related weather events, the data they relate to and a stark warning about our future through the mind of a machine. It is part of the artwork series Heartbeat of the Earth launched in collaboration with the UNFCCC.

the artwork uses the metaphor of an MRI to scan the Earth for damages inflicted by human activity over the last 50 years, and to visualise how extreme climate events are connected to climate change data.

You can launch the experiment here.

By using visual cues to understand the invisible numbers around us, we are invited to learn about how the planet's stability has been affected. Major climate related weather events from 1970 to present are highlighted in a timeline, with data on population, temperature and CO2.

This is one of the largest datesets ever collected of nature imagery. Dive into a collective memory of more than 200 million images of landscapes taken by the human lens.

Training an AI model on this dataset, the machine learns from our memories, and dreams up artificial morphing visualisations showing the Earth’s beauty. A hint that if we fail to listen to the data, these might be the last memories we have of nature: digital representations.

The experiment uses a Machine Learning model trained on 200 million images of Earth's landscapes, from satellite data and found open source imagery, to generate GANs (Generative Adversarial Networks) that depict a moving visual of Earth’s beauty dreamt by the machine.

The artwork contains one of the largest datasets of open source nature imagery with over 200,000,000 images of the Earth and its natural wonders, acting as a collective human memory of the beauty of our planet.