Deploying AI algorithms in the fight against human-caused extinctions

Humans are spurring extinction at a horrifying rate, but AI algorithms may just be the solution, according to S. Anand, co-founder and CEO of Gramener


Last year, the WWF released one of the most shocking statistics of all-time: Humanity has wiped out 60% of the world's mammals, birds, fish and reptiles since 1970, casting a gruesomely stark light on the damage we as a species are doing to our world.

"We are sleepwalking toward the edge of a cliff," said Mike Barrett, executive director of science and conservation at WWF. "If there was a 60% decline in the human population, that would be equivalent to emptying North America, South America, Africa, Europe, China and Oceania. That is the scale of what we have done."

You are not alone in feeling hopeless in the face of such devastating facts, especially in a world where the innovation driving us forward is also a big part of what is causing the environmental problems that threaten to destroy us. However, innovation may have just given us our best ally in the fight against mass extinction: AI.

DATAx spoke with S. Anand, co-founder and CEO of Gramener, a data science company whose work with a number of environmental agencies led it to win the CNBC–Microsoft award for "Empowering Societal Progress with AI", about tackling extinction using AI algorithms for data analysis.

Gramener's work

"Our company brings data to life through insights, for which we apply AI and ML," explains Anand. "We work with clients with known data and known problems providing them with data science expertise."

Gramener's aim is to help people actually understand their data so they can put it to practical use.

"We work with organizations across industries, including NGOs, social organizations and governments, as well as public-facing organizations and a variety of stakeholders," Anand tells us. "Our aim is to simplify data so people get the insights they need."

AI and fish migration

Anand first tells us about the company's partnership with Microsoft AI for Earth working with Nisqually River Foundation to monitor fish populations.

The Foundation had installed a camera and infrared sensors in a fish ladder at a dam on the river which was triggered for capture 30 seconds of video whenever a fish swum past the sensors. An expert would then analyze the video to discover what kind of fish it was to gain insights into the migration patterns of the different species.

"This is important because we want to maintain a certain amount of ecological balance and key to this is understanding the migration patterns of different species of fish," Anand explains. "The trouble is that it is a nightmare to track manually for a couple of reasons. First off, it requires someone to view hours upon hours of video and secondly it requires an expert who is reasonably experienced.

"So, we put an AI application in the camera pre-trained by biologists on images from a variety of species of fish, that way the expertise of these biologists was transferred into this AI. As a result, it has an accuracy on par with the best marine biologists. Video footage that would have taken a few experts several months to sift through is now entirely automated," he says.

Using the deep learning and AI models, Gramener and Microsoft were able to ensure fish identification of 73% accuracy.

Plant species identification using AI algorithms

Anand then talks to us about a client the company worked with to identify a number of plant species

"Again, the problem was the same in that it takes an expert a fair amount of time to actually identify the species," he says. "But when you train an AI model with the expertise it becomes equal to an expert system and is able to work quicker and more efficiently."

AI identifying tree maturity

"We've also worked with a plantation company in Southeast Asia who want to ensure they are deforesting responsibly," Anand says. "They want to make sure they aren't cutting down trees which are not mature and therefore unsustainable for harvesting. The way they do this is by using aerial photography to identify the age and development of the trees in certain areas to ensure they are large enough that they cannot grow any bigger, making deforestation both ecologically and economically viable."

In steps Gramener. Its AI system is specifically created to detect this sort of pattern from images.

"The system is now automated," Anand tells us. "This is not something that necessarily requires an expert, but the area of coverage is so large that automation makes a huge difference in this case."

Identifying proteins using AI

Gramener's work saving the planet does not stop at identifying the species that walk, swim, grow or fly on its surface, it is also able to identify things on a molecular level.

"We've also been working with a pharmaceutical company to identify proteins present in a formula after it's been exposed to a disease," says Anand. "When experts identify the formula, they can see the protein shapes and that way – and this sounds a bit silly and simple I know – they can identify it by whether it's a "rectangle" or a "triangle", for example.

"However, one image can have thousands upon thousands of these shapes, and it takes an expert who knows what the protein looks like an incredibly long time to chart these shapes. It's an extremely time-consuming and dull process for a human, but AI can map these shapes much quicker and achieve a good level of accuracy using the expert's knowledge," he adds.

I see… patterns

When it comes to AI's place in battling against mass extinction, evidently it is a case of distinguishing patterns in images.

"All of these solutions are related to recognizing patterns that an expert is able to do but a non-expert is not," concludes Anand. "And there's a widespread need for this sort of approach."

For Anand, the problem of extinction comes down to one question: "Can we efficiently automate what a human expert would take a long time to do?"

Blockchain is set to solve the fake food problem small

Read next:

Blockchain is set to solve the fake food problem