AI Is Discovering New Species Quickly, But Are The Results Trustworthy?


Artificial intelligence (AI) is changing the game for scientists studying wildlife and discovering new species.

But how reliable are these new methods?

Currently, AI helps by flagging potential new species, which expert biologists must then formally describe and classify.

However, AI’s effectiveness depends heavily on the quality of the data it’s trained on, and there’s still a lot we don’t know about Earth’s wildlife.

AI is valuable for making sense of large datasets collected through smartphones, camera traps, and automated monitoring systems.

“We’re speeding up research to tackle bigger questions, which is exciting,” says Christine Picard, a biology professor at Indiana University.

Success Stories and Challenges

In 2023, Picard and colleagues published a study in Methods in Ecology and Evolution.

They trained an AI model to identify over 1,000 insect species using images and DNA data.

The AI correctly identified 96.66% of known species and placed unknown species in the correct genus 81.39% of the time.

However, when relying on images alone, accuracy dropped to about 39%.

The low accuracy was partly due to the poor quality of images from public databases.

“Some photos were quite bad, so it’s impressive the model did as well as it did,” Picard noted.

Exploring Biodiversity

AI is especially useful in studying biodiversity, particularly in the tropics where most of Earth’s species are found, but research is scarce.

Insects, the most diverse animal group, are largely unidentified.

AI can help bridge this knowledge gap.

“It lets us explore the unknown diversity of insect species,” Picard says.

AI is also used to monitor ecosystems.

Jenna Lawson from the UK Centre for Ecology and Hydrology uses automated monitoring systems (AMI) equipped with cameras and audio recorders to track species.

These systems gather more data than humans can process without AI.

“We have amazing hardware to collect data, but without AI, we can’t analyze it,” Lawson explains.

Katriona Goldmann from The Alan Turing Institute is training AI models to identify animals from these systems.

Her models can alert researchers to unknown species, flagging unusual images.

Monitoring Biodiversity Changes

AMI systems help track biodiversity over time, crucial for conservation.

Human activities are driving species to extinction 100 to 1,000 times faster than normal.

Monitoring wildlife helps us understand and respond to environmental changes, including climate and human activities like agriculture.

“Technology in biodiversity research is fascinating; it lets us record data at an unprecedented scale,” Lawson says.

However, AI systems are energy-intensive, which can impact the environment.

To address this, Goldmann is compressing her models to run on small, solar-powered computers attached to AMI units.

In conclusion, while AI shows great promise in identifying new species and monitoring biodiversity, it’s still a tool that needs refinement and expert oversight.

With continued improvements and responsible use, AI can significantly enhance our understanding of the natural world.