Identifying Life and Non-Life through time and space
Understanding the differences between Life and non-Life using bottom up approaches is the goal of my research field, Artificial Life. I try to achieve this goal by using both AI-based and purely statistical tools that can be applied to either synthetic or natural data (also called “substrate-agnostic” tools). We only know one example of Life, a serious limitation for any scientific study. Overcoming this issue could come through (1) the search for other forms of Life with an independent origin from our own or (2) the fully artificial creation of a different form of Life, in simulation or in vitro. Two of my research topics are therefore Astrobiology and Open Ended Evolution.
Our search for Life in the Universe is necessarily biased by our knowledge of the biology of Earth-based Life. Agnostic biosignatures, signs of Life that do not rely on biological expectations, are one way to sidestep these biases. In 2022, I co-authored a proposal  for a new kind of agnostic biosignature: a complexity-based method to find signs of Life in the universe.
We showed that a measure of the complexity of electromagnetic time-series data from real and simulated planets is correlated to the diversity of the type of surfaces on that planet. In other words, the patterns of light from a distant planet, even when the planet is so far that its size is reduced to one pixel, can tell us if that planet has a combination of various types of surfaces (in the case of Earth, oceans, forests, clouds, deserts...). I am currently working on other types of agnostic biosignatures.
Figure 1. The diversity of surfaces of a planet can be approximated by the statistical complexity and the Shannon entropy of the electromagnetic time series from that planet, even at large distances. From .
Living systems, including individuals but also societies and their technological advances, are the only known examples of systems that seem to evolve towards ever more complexity and novelty. This property is called Open Ended Evolution. To understand Open Ended Evolution, I find necessary to study both the origins of life  and its dynamics through time. I currently research these areas through evolutionary simulations. In 2023, I published a paper demonstrating exponential genetic drift in an AI-based cellular automaton , a first step towards Open Ended Evolution in a fully synthetic system.
Figure 2. A multi-agent simulation showing different origins of life and the diversification of individuals into many different species over time (x axis). The y and z axes, as well as the colors, represent different characteristics of each individual such as the size, speed etc. Unpublished.
Figure 3. A cellular automaton where generations exponentially diverge both genetically and phenotypically. From .