Skip to Main Content
Skip to main content
Microsoft News Centre UK
man works with technology in a field

‘We need to do some radical things and we need to do them now’

Microsoft’s Chief Environmental Officer reveals why he thinks the world of AI for Earth

On October 8 last year, the United Nations published a report that called for global warming to be limited to 1.5 degrees centigrade over the next 12 years. Failure to do so will significantly worsen the risk of drought, floods and poverty for hundreds of millions of people, scientists warned.

The research made for grim reading and laid bare the challenge that mankind faces in creating a healthy and thriving planet that future generations can live on.

Lucas Joppa, Microsoft’s Chief Environmental Officer and the man behind the company’s $50m AI for Earth programme, is honestly blunt when asked about the UN’s findings during a visit to London recently.

“There are two conclusions you can take from the report,” he says. “One is we are finished; but I’m not a fatalist, so I try not to take that route. If you reject that conclusion, which I hope human society does, you are left with only one other – we need to do some pretty radical things, and we need to do them now.”

Three weeks later, the WWF published a report stating that global wildlife populations have fallen 60% since 1970.

Decades of climate change, pollution and the overuse of natural resources led the conservation organisation to conclude that “the variety of life on Earth and wildlife populations is disappearing fast”. From a financial perspective, economic losses in the US alone from extreme weather and the health costs of air pollution will hit $360 billion annually in the coming decade, according to a report by the Universal Ecological Fund.

Those are big statements. The even bigger question is who can solve what many consider to be the greatest crisis the world has ever faced?

“It requires everybody to lean in, and some will have to play almost disproportionate roles,” Joppa says. “Governments need to do their part and every person has to do their bit. But the tech space has a major role to play in deploying technologies, human resources and expertise. We have a lot to do and not a lot of time to do it in.”

With an announcement of a bolder ambition from Microsoft President Brad Smith, calling for a tech-first approach to sustainability and the embedding of sustainability as a core value across all business units, it’s clear that Joppa is far from alone in that belief at Microsoft.

Even if you somehow manage to brush off the potentially catastrophic UN and WWF reports, you can’t ignore Joppa’s first-hand experience and knowledge in the environment sector. He holds a degree in Wildlife Ecology and Zoology from the University of Wisconsin-Madison – close to where he grew up – and a PhD in Ecology from Duke University; he’s advised the UN, been a member of the Science Advisory Board at Natural England, is an Honorary Conservation Fellow at the Zoological Society of London, and has sat on numerous boards, including the Federal Advisory Committee for the Sustained National Climate Assessment in the United States. He also spent time in Malawi, volunteering with the US Peace Corps.

Joppa joined Microsoft in 2009 as a Computation Ecologist, based at the company’s Research Lab in Cambridge. Five years later he moved to Microsoft’s global headquarters in Redmond, in the US, and took on his current role in July after creating the AI for Earth initiative that was launched by Brad Smith, company President, last year. It’s a new position for him, and Microsoft, which shows how seriously the company is taking the issue of climate change.

There are two conclusions you can take from the report. One is we are finished; but I’m not a fatalist, so I try not to take that route. If you reject that conclusion, which I hope human society does, you are left with only one other – we need to do some pretty radical things, and we need to do them now.

Joppa’s aims are clear: “We ask ourselves: ‘Are we doing what is needed to address the defining environmental issues of our time? Are we driving down emissions and our negative environmental impacts? Are we using our buying power and partnerships and technology to accelerate positive environmental change outside of our four walls? And we don’t stop until the answer is yes.”

Those are daunting objectives, even for one of the largest technology companies in the world, but Joppa is fully aware of the task ahead. After moving back to the US he wrote a memo outlining how cutting-edge technology could be used to tackle some of society’s most pressing problems. That idea grew into AI for Earth – a programme offering cloud and AI computing resources, training and grants to researchers across the world who are trying to create a more sustainable future.

The demand for that programme was so big that an initial $2m, one-year commitment that was unveiled in London on July 12, 2017 turned into a $50m, five-year programme announced at the One Planet Summit in Paris in December 2017.

“We did that [launch with $2m] to see if our intuition was correct, which was there was going to be a lot of demand for this,” Joppa says. “The response in the first couple of months was extraordinary, and it was coming from all over the world, from all sectors. People were excited and wanting to get involved.”

More than 230 grantees in over 60 countries have received grants so far, covering every continent – a feat Joppa calls “truly extraordinary”.

Seven of those organisations are in the UK, and some of the projects include: in Cornwall, AI is being used to identify seals; in Shropshire, one company is deploying machine learning to understand the perfect time to pick coffee beans across the world; the Royal Society for the Protection of Birds is monitoring wildlife in Sierra Leone and Liberia; and The University of Edinburgh is using Microsoft’s cloud platform to help animal researchers and volunteers communicate more effectively.

Joppa believes the UK is “an exceptionally special place” for research because of the concentration of talent, enterprise and education in such a relatively small area. It also helps that the UK has a history of exploration, a juxtaposition of traditionalism and modernism, he adds, the clash of the natural world and cutting-edge technology.

The National Oceanography Centre in Liverpool is using a grant to try to predict wave sea states in the North Atlantic by using deep learning.

Dr Nicolas Bruneau, a Scientist at the organisation, said: “Wave dynamics are a key part of our global Earth system; however, current global climate models don’t directly take waves into account as solving them deterministically is complex and computationally expensive.

The National Oceanography Centre in Liverpool is using a grant to try to predict wave sea states in the North Atlantic by using deep learning

“My project focuses on developing a deep‐learning framework to emulate key wave characteristics (necessary to account for wave interactions with the ocean and atmosphere) in an accurate and cost‐effective way by reducing the number of calculations once the machine learning model is trained with a large amount of data.”

He added that Microsoft’s computing power would provide huge benefits to his study.

“The AI for Earth programme allows easy access to powerful, cloud‐based computers. It’s a great way for scientists to access the Microsoft Azure platform and experiment with it to assess its potential for future use. I knew immediately that it could provide access to new infrastructure that was not yet available at the National Oceanography Centre.”

All the grant recipients are building on Microsoft’s 35-year work with AI, which aims to “assist humanity and augment our capabilities”, according to Harry Shum, Executive Vice-President of AI and Research at Microsoft. As a result of work at sites such as the Cambridge Research Lab, Microsoft is currently one of the world’s leading experts in AI technology.

“People forget that Microsoft has been investing in AI research for the past 35 years; this isn’t new for us,” Joppa says. “AI for Earth represented the first cross-company effort to deploy those 35 years of investment in research and technology in a key area of societal importance. It’s focused on agriculture, water, biodiversity and climate change, and acknowledges the fact that, as a global society, we have to figure out how to mitigate and adapt to changing climates, ensure clean water supplies, sustainably feed people around the world, and stem the global loss of biodiversity. We have to tackle all that together, and in the face of growing human populations. I’m not that old, and there are twice as many people on planet Earth as the year I was born. That growth is expected to continue on up to 10 billion people.”

Joppa points out that one of the few things matching the exponential rise of negative human influence on the planet is the development of technology and innovation.

“We think there is going to be a significant role to play for AI in general and machine learning in particular in building solutions there,” he adds. “The AI for Earth grantees are making significant progress and I’m excited about the growing, global nature of the programme.”

The Committee on Climate Change agrees that it will take more than legislation to address the problem, and “will involve a combination of new technologies, processes and human behaviour”.

More than 230 grantees in over 60 countries have received grants so far

As such, Microsoft is uniquely positioned to help; it has the staff, the set-up, the reach and the financial strength. It is also leading by example by being carbon-neutral since 2012, putting a formal price on carbon and building a net-zero-water campus in California. Joppa believes the company’s involvement could be a game-changer.

“One of the great ironies of our day is that we often ask the organisations with the fewest resources to do the most – environmental non-profits, academics, under-funded governmental agencies. They’re the ones tasked with solving one of humanity’s greatest ever challenges?! That’s absurd – we need absolutely everyone leaning in right now.

“I want to make sure we are able to give the grants to organisations, whoever they are, as long as they are taking a machine-learning-first approach.”

However, simply giving organisations access to cutting-edge technology is not enough; the people who work for those institutions need to know how to use it effectively. Environmental experts are not computer scientists; they will need training and education programmes to attain the skills needed to help them with their projects. Those experts might know how to log data, collate it and study it, but they may not know how to interpret huge amounts of data that have been subjected to Microsoft’s powerful AI algorithms.

“Just because you give people technology, it does not mean they know how to use it,” Joppa says. “Everyone is an expert in something, so how do we empower people to take what they are good at and use that in the world of environmental sustainability, with AI accelerating their contribution. What are the education and skill programmes we need to deploy to ensure that those people who run environmental non-profits or are environmental faculty members or work in these government agencies … we recognise that they were often not the ones getting their degrees in computer science. But if we want them to be able to use our technology as effectively as possible, we need to help them get those skills.”

Microsoft has put together an AI for Earth team that is helping grantees and partners across the world to build the applications that can help the planet. The aim is to create a connected and seamless digital world that can be used by everyone involved in Microsoft’s programme, whether they are just about to apply for a grant in the US or conducting experiments in a field in Africa.

One example is iNaturalist – a company that encourages everyone to be a naturalist. The app lets people take photos and videos of animals and plants and upload them to a database, where scientists can use that information to learn more about the environment across the world. The leading contributors have logged thousands of observations of hundreds of different species in November alone.

iNaturalist lets people take photos and videos of animals and plants and upload them to a database Photo credit: iNaturalist

“iNaturalist was one of the first grants we handed out and we’ve only deepened our investment with them since,” Joppa says. “They highlighted the same approach that we have to the deployment of AI, which is a recognition that technology doesn’t solve human problems; humans solve human problems. Technology can make humans much more efficient, much quicker and with greater expertise.”

Imagine you are on a hike and you spot a butterfly. You take a photo of that animal with iNaturalist and that image is sent to a database, where experts can identify it. You can also chat with those experts about where you saw the butterfly and what time of day and year it was, to build up a picture of the local wildlife. Now imagine that is happening in countries across the world, and you suddenly have a network of amateur naturalists who are collecting data and doing the work of expert scientists.

iNaturalist is also using technology to help its community become better at identifying those creatures.

“If you have a photo of a species and show it to a computer vision algorithm, it will tell you which species that is around 65% to 70% of the time,” Joppa says. “If you ask the algorithm for the top five most likely species, it jumps to about 95% accuracy. You don’t have to be a taxonomy expert to look at the species you saw, look at five photos and match them. That’s what iNaturalist did, and we worked with them to scale their app on Azure.”

Microsoft has a range of powerful image recognition programs in its Cognitive Services suite, which are used by companies to improve how they work with customers. However, none is operating in the AI for Earth programme yet. Joppa is trying to change that.

“Once you have that baked into the app, you can start doing other things with machine learning, such as improve the ability of the classification algorithm,” he says. “It could tell me the probability of any particular species occurring in a place based on temperature, humidity or elevation. Then you would have a probability of what a species is based on what it looks like and where you found it. Put those together and your accuracy goes up significantly. And we could start to incentivise the people using the app to go and look in certain places for animals and plants. Right now, it’s a very passive thing.

“Machine learning can play a critical role when you are taking novices and increasing their level of expertise so they can be more efficient in doing a task that the world desperately needs done.”

Just because you give people technology, it does not mean they know how to use it… if we want them to be able to use our technology as effectively as possible, we need to help them get those skills.

There are few more important goals for mankind than tackling the diseases that can kill huge numbers of people every year. According to the latest figures from the World Health Organisation, 86 countries and territories have reported evidence of mosquito-transmitted Zika infection, including a large outbreak in Brazil in 2015.

Meanwhile, there were 216 million cases of malaria in 2016 worldwide, with 90% occurring in Africa. These cases resulted in an estimated 445,000 deaths, just 1,000 lower than the previous year.

For Project Premonition, Microsoft built the world’s most advanced mosquito trap

Tackling both diseases has proved difficult due to a lack of funding in affected countries, risks to humans in endemic zones and anomalous climate patterns. One of Joppa’s favourite AI for Earth projects – Project Premonition – could be the answer.

“Why are major diseases so catastrophic? It’s because we have no idea that they are coming; they take us by surprise, and humans struggle with that,” he says. “To stop that happening you have to monitor those diseases before they break out in populations, but that means you have to track the disease while it’s still in wildlife, often in remote environments.”

In echoes of Jurassic Park, Joppa says the easiest way to obtain those biological samples is by capturing mosquitos, which are “syringes on wings”. To do that, Microsoft built the world’s most advanced mosquito trap.

“We built devices that are running machine learning algorithms and can be deployed in the wild. Any time an insect passes through the high-frequency LED, it casts a wingbeat shadow,” Joppa says. “The device analyses that shadow and classifies the insect down to the species, determines whether it should capture that insect and shut a door to capture it – all in a few milliseconds.

“We can quickly deploy the device and bring it back to the lab, convert the analogue data [the genetic information) to digital data [the genome sequence], put that through a metagenomics pipeline and match that up with all the lifeforms on Earth. We can then create a Power BI dashboard of all the diseases and species in that ecosystem.

“We’ve come a long way since we started that project, and we’re still a long way from its full potential, but it could be truly transformative for how humans monitor the Earth’s natural systems.”

Keeping a close eye on how humans are affecting the world is best achieved through analysing data. “Computational approaches are the defining technology of our time,” Joppa adds.

With AI for Earth, Microsoft hopes it can empower organisations to collect enough data to have a meaningful impact on improving the world. Joppa is clear that Microsoft will not own the data, it will only help to store and analyse that information. “Trust is built, earned and given,” he says. “It’s all about the choices you, as a company, decide to make.

“Microsoft just builds the technology and services to make our partners better. That’s the model we are deploying in AI for Earth.”