Dr. Lucas Joppa grew up in the wilds of Wisconsin, fascinated with the many species that roamed the national forests. He graduated college with degrees in Wildlife Ecology and Zoology and spent the next two years working in wildlife conservation as a Peace Corps volunteer in Malawi.
He probably didn’t envision a career at Microsoft, but life has a funny way of revealing solutions. Today, spearheading Microsoft’s AI for Earth Project is Joppa’s primary occupation. He spent 2017 dedicated to delivering a single message: the planet needs Artificial Intelligence (AI).
The mention of AI often brings up images of humanoid robots doing our chores for us - shopping or mowing lawns. However, the more exciting AI advances instead highlight how computers are unlike human beings. Machine Learning is case in point. Although the idea stems from a very human attribute, when applied to machines learning is a superhuman trait.
People learn slowly. Even the cleverest humans need time to process and incorporate new information. Our trials and errors (data points) become lessons over time. The time involved in machine learning is exponentially smaller. The number of data points that a machine can reliably process and incorporate are huge, and ever expanding. The creation and evolution of big data applications for example, are in their infancy, but it is a phenomena already driving total industrial change.
Big data and machine learning are the centre of widely publicised developments such as self-driving cars and digital marketing campaigns, but they can also be applied to environmental causes. It is this game-altering possibility that led Joppa to Microsoft. As a child counting sightings of animals in Wisconsin, he became fascinated with wildlife. As a well-educated and well-travelled conservationist, he became painfully aware of threats to biodiversity. As Joppa points out in a 2016 Science editorial, species are disappearing at a rate 1000 times the expected background level of extinction and we can expect this rate to rise to 10,000 times the norm unless something is done. The “biological annihilation” of wildlife over the past 100 years means a sixth mass extinction in Earth’s history is in process and is far more severe than previously feared. Having analysed both common and rare species, scientists have found billions of regional or local populations have already been lost due to human overpopulation and overconsumption. The implications of this mass extinction, termed the Sixth Great Extinction, is a threat to the survival of human civilisation, unless we act quickly.
How can AI help?
eBird provides a glimpse of what is possible. A crowd-sourced, citizen-scientist platform intent on collecting the data of the birds, eBird was launched in 2002 by the Cornell Lab of Ornithology and National Audubon Society. Bird watchers from across the globe are encouraged to submit their observations in a user-friendly format that can be easily organised, and analysed by smart machines. The data is open source, free to all without guidelines, requirements, or strings attached.
As of 2017, researchers had identified 159 conservation efforts born of the eBird database. The specifics are varied, but the result is the same: big data is making a difference. A review of how birds were using the Fraser River Delta convinced the Canadian Government to protect the area as a Wetland of International Importance.
At National Wildlife Refuge Centers, managers use eBird data to implement the most effective wetland impoundment schemes in support of birds’ habitats. None of this would be possible without the AI that supports and processes eBird data.
The potential applications that a worldwide, high resolution, satellite-based system of data collection could bring are only beginning to be realised. Planet, a Silicon Valley company aiming to image the entire Earth in order to spot trends and collect data to make global change visible, accessible, and actionable, has launched enough satellites to get the ball rolling. They use algorithms and machine learning to analyse and understand the immense data being collected - about one and a half million images per day, every day. They intend to be for the physical world, what google is for the internet. Users will be able to use this information to assess the damage of natural disasters, measure urban sprawl, or track deforestation.
AI is also being used in energy efficiency measures, agriculture and conservation efforts, worldwide. For example, smart grids are can minimise waste by allocating power using two-way communication within the system and advanced algorithms. Genomics, unmanned aircraft systems, and satellite technology are linking arms with big data in agricultural management in an attempt to grow better crops in an environmentally friendly manner. Environmental models, though far from perfect, allow researchers to change inputs and receive new outcomes quickly and without additional work. As we improve these models with ever increasing AI algorithms, we will also improve our resource conservation efforts.
We are at the beginning of our understanding in how we can apply AI to aid global sustainability efforts. In a surprising twist, we may not be the designers for much longer. Google, for example, has been automating the system via AI, to build sophisticated machine learning AI. In 2017, AutoML was incorporated into the design process, creating NASnet, a child machine learning AI that outperformed its man-made equivalents in every task assigned. If you’re suddenly thinking about Skynet and the machine take-over, you’re not alone. As the late Stephen Hawkings stated, AI will be “either the best, or the worst thing, ever to happen to humanity.” With our planet hanging in the balance, the cost of not developing AI may be too great than the risks that accompany entering the new paradigm of ‘thinking machines’.
If you have any questions
