4 April 2024

11:45 – 12:30

Amphi 3

Roman Sarrazin-Gendron

Roman Sarrazin-Gendron is a lecturer at UQAM and a doctoral candidate in computer science at McGill University as part of the Games for Good initiative, dedicated to the scientific design of responsible video games. Trained as a bioinformatician, he is developing new interdisciplinary methods for analyzing biological data, combining biology, computer science and cognitive science. His doctoral thesis focused on Borderlands Science, a participatory science game released in 2020 that mobilized millions of players to align the DNA sequences of the human microbiome.

Can video games provide the human resources needed to analyze scientific data?

Every day, scientific equipment, from telescopes to microscopes, captures millions of gigabytes of data. This deluge of data, which needs to be classified, cleaned and analyzed, far exceeds the technical and human capacities of the scientific community. The rapid development of artificial intelligence (AI) technologies offers a possible solution, with excellent results on many types of data, as seen with ChatGPT and text. However, not all data lends itself equally well. For example, the heterogeneity of DNA sequences makes it difficult to train algorithms to align them, and artificial intelligences still make mistakes in this type of task, which is essential for identifying disease-related mutations and understanding the evolution of species. So what to do when AI is no longer enough? The answer may lie in video games. Every week, humans collectively invest hundreds of thousands of years solving complex virtual problems. What if this energy could be channelled into solving scientific problems? That’s the challenge of the Borderlands Science video game, which has mobilized over four million players to align DNA sequences in order to improve the classification of human gut microbes.