Artificial intelligence helps insects – 11/21/2024 at 08:15

What if artificial intelligence could be an ally in helping insects? This is the bet made by a team of Montreal entomologists supported by engineers, to document the unprecedented decline of these millions of species and better counter it.

Under the large transparent dome, thousands of butterflies of all colors, chrysalids, ants and praying mantises… It is here at the Montreal Insectarium that this initiative was thought up, notably by its director Maxim Arrival.

Compared to “all the mass extinctions that we have experienced in the past, the one that affects insects is happening a thousand times faster,” explains the enthusiast.

And even so quickly “that we are unable to follow it adequately to put in place the necessary actions to slow it down”, adds the entomologist.

Disappearance of habitats, pesticides and climate change: if the reasons for the disappearance are known, we have little data on the exact scale of this massacre.

A gap that the Antenna project would like to fill, which uses AI and therefore an algorithm that identifies insects using photos.

Everything is powered by solar terminals, placed in the Canadian Far North but also in the Panamanian tropical forests. They are designed to take a photo every 10 seconds of insects attracted to UV light.

Researchers estimate that this innovation will double the amount of information on biodiversity collected over the last 150 years in two to five years.

“Even for us, it looks like science fiction,” adds Maxim Larrivee, with a smile on his face.

Ultimately, this data should make it possible to create “decision support tools for governments and ecologists”, to identify the best conservation programs to adopt and “restore biodiversity”.

– Next major breakthrough –

Often little known, insects represent half of the world’s biodiversity and play a crucial role in the balance of nature, whether through pollination, the transformation of waste into fertilizer or by constituting the basis of the food chain of many animals. .

“This is the next major advance in the field of biodiversity observation,” thinks David Rolnick, researcher at Mila, the artificial intelligence institute of Quebec.

The innovation has been in the testing phase for several weeks, the model is “open source” and is currently only focused on moths.

With more than 160,000 different species, they represent an “extremely diverse” group, “easy to identify visually” and they constitute the “base of the food chain”, specifies David Rolnick, expert in AI and passionate about insects since his earliest days. young age.

Ultimately, the idea is to allow everyone to contribute to enriching the platform, but also to train the AI ​​to recognize new species of insects. Because if more than a million are already known, there could be ten times more.

“It is estimated that 90% of insects have not yet been identified by scientists,” explains the researcher.

In one week, a station installed in the Panamanian jungle discovered “three hundred new species”, specifies David Rolnick, adding that “this is only the tip of the iceberg”.

The researchers also hope to be able to use this computer model to identify new species in the deep sea, or even harmful species in agriculture.

In Montreal, the Insectarium already uses technology for educational purposes.

Visitors to this museum dedicated to insects can take photos of butterflies roaming free in a vivarium and find out what species they are via the app.

It’s a discovery for Camille Clément, a French tourist: “Artificial intelligence in the service of ecology is a good thing, if we use it carefully,” she believes while photographing the butterflies that fly around her.

For Julie Jodoin, director of Space for Life, a group of five Montreal museums including the Insectarium: “If we don’t know nature, we can’t ask citizens to change their behavior.”