A collaboration project with AI motion graphics designer Kristaps Kazaks to explore the possibilities of the cutting-edge AI text-to-image generation technology and what it could tell us about nature.
By taking inspiration from one of the longest scientific experiments in history which shows that chaos and unpredictability are essential parts of nature, the project uses AI to visualise and predict it.
What would it mean to accept both nature and AI as unpredictable entities? Would one eventually help to predict the other?
By taking inspiration from one of the longest scientific experiments in history which shows that chaos and unpredictability are essential parts of nature, the project uses AI to visualise and predict it.
What would it mean to accept both nature and AI as unpredictable entities? Would one eventually help to predict the other?
In the spring of 1989 a sample of water from the Baltic sea was put into an outdoor experimental system. All of its natural microbial life and its processes were observed and sampled twice a week, for 2319 days, until the autumn of 1995.
The natural interactions between the micro-organisms generated chaos - the ecosystem surprisingly never reached a stable state with long-term predictable food chains. The predictability for this experiment turned out to be a small time horizon of 15–30 days, only slightly longer than the local weather forecast.
This shows that stability is not required for the existence of complex food webs and suggests that the long-term prediction of future for any species can be fundamentally impossible and that nature is governed by more chaos than we think.
This project uses cutting-edge AI text-to-image generation to create animations based on familiar verbal concepts. Similar to the 1989 experiment, the AI is given text input of natural phenomena and asked to generate a moving image from it. The resulting animations are full of unpredictability and chaos that strangely resemble the one observed in nature.
What would it mean to accept both nature and AI as unpredictable entities? Would one eventually help to predict the other?