Oliver Balch reports on how the tech giant’s first ever chief environmental officer is aiming to ‘optimise a super-simple equation’ – reduce Microsoft's carbon footprint and maximise its positive impact by harnessing the power of AI
For a man who grew up in the back woods of Wisconsin with no television nor computer, Microsoft might seem like an odd career choice.
Lucas Joppa makes no effort to hide the fact. He never intended to “land in a large corporation”, he says. His purpose, instead, has always been to “seek the greatest kind of environmental additionality possible”.
That sense of personal mission is paying off. In July 2018, 37-year-old Joppa was appointed Microsoft’s first ever chief environmental officer. He describes the role as dual-focused: part internal (i.e. impact minimisation) and part external (i.e. maximising potential upsides of his employer’s tech).
Even if we do reduce all of these things to zero, it's far from enough for the world to achieve what it needs to achieve
“I talk about my team just everyday working to optimise a super-simple equation, which is minimising the environmental footprint of our business operations and maximising the positive environmental impacts of our products and our policies and our partnerships.”
The first half of the job is far from small fry. The carbon emissions created from Microsoft’s own operations, coupled with the effects of the goods and services it buys and sells, come to just shy of 23 million metric tonnes – similar to the total carbon footprint of Sri Lanka.
The tech giant has a long way to go to bring this number down, but Joppa takes heart from the fact that he isn’t starting from scratch. Since 2012, for instance, the company has achieved carbon neutrality by introducing an internal carbon fee – the price of which, incidentally, it nearly doubled (to $15 per metric tonne) back in April.
Joppa insists that Microsoft is “in no way stepping down” its commitment to operating in as environmentally sustainable manner as possible. Yet, at the same time, he concedes that the impacts of Microsoft’s server centres, PCs and so on remain “a drop in the bucket” when compared with the colossal volume of greenhouse gas emissions globally.
The conclusion is unavoidable: “Even if we do reduce all of these things to zero, it's far from enough for the world to achieve what it needs to achieve.”
Conveniently, his logic here appears to coincide with the half of his job that really gets Joppa’s pulse racing: namely, putting Microsoft’s vast technological and intellectual weight behind breakthrough solutions to pressing environmental challenges.
Less than three years in and AI for Earth has already backed over 380 projects
And it is this that has caused Joppa’s star to shine so brightly at Microsoft in recent years. In early 2016, in his then capacity as a computational ecologist, he wrote a short internal memo outlining how the corporation’s artificial intelligence resources might be put to the service of eco-entrepreneurs worldwide.
From the back-of-an-envelope idea, the concept grew fast, catching the interest of colleagues and quickly winning the ear of senior management. One year later and AI for Earth was born.
The initiative was initially pitched as a one-year, $2m programme providing peer technology grants. Such was the avalanche of early responses that Microsoft promptly ploughed a further $48m into the project and expanded its timeline to five years.
The programme is premised on three pillars, Joppa explains: access (to Microsoft’s tech), education (in computer science wizardry), and talent (i.e. drawing on Microsoft’s internal, techie brain pool).
“The pillars of the programme were predicated on these areas that we knew would have to be solved if we were going to allow people in the sustainability and the environmental space more broadly to grow into the AI space,” Joppa explains.
The combination appears to be working. Less than three years in and AI for Earth has already backed over 380 projects. Most of the successful candidates are US-based, although the programme’s total reach currently stands at 66 countries from every corner of the globe.
An example like iNaturalist really just shows me what’s possible with a small organisation with access to global-scale technology
All the projects follow a similar technology pipeline to Microsoft’s conventional development pattern. At the heart of all the AI for Earth innovations is the deployment of algorithms as application programming interfaces – referred to in tech lingo as APIs. These APIs go on to form the basis of end-of-user applications that are hosted in the cloud and made available on a range of different types of devices.
“Once they [grantees] have that model developed, there's an easy way for them to deploy the raw file of code into a web-based service and allow anybody anywhere around the world to access it,” says Joppa.
To illustrate what he means, he cites an app called iNaturalist. Developed by National Geographic and the California Academy of Sciences, the app allows everyday folk to become citizen scientists by allowing them to upload a photo of any type of living organism and obtain an identification of what it is.
The identification process works by taking expert information about a raft of different species and encapsulating it in a deep neural network that then makes the match. It helps, of course, if users can provide some basic taxonomical knowledge along with the original photo. Failing that, however, the app provides a series of flip cards to narrow down what the answer might be.
The sense of excitement in Joppa’s voice as he describes the solution is inescapable. The small team behind the project has now launched a live augmented reality version of the app, called “Seek”, that identifies and names species in real time as users video their surroundings.
“An example like that really just shows me – and I hope it shows the world – what’s possible with a small organisation with access to global-scale technology,” he enthuses.
Stuff like that kind of blows my mind. It was the first time we actually even asked the question and got an answer to, ‘Where are all the trees in the United States?'
Joppa has a hatful of similar examples. Another that particularly excites him is called Wild Me, an app that scours social media platforms such as Flickr and Instagram for images of individual species and assigns them an identifiable name. Plug this data into a large-scale modelling algorithm and it suddenly becomes possible to create far more accurate global wildlife population assessments.
Before he wraps up, Joppa cites one more stellar grantee, SilviaTerra. A US forest mapping enterprise, it applied to AI for Earth to build up an algorithm that extrapolates high-resolution satellite imagery to determine information about the health, volume and location of trees in specific areas.
The firm recently completed the first ever forest inventory analysis for the whole of the US. Using over 800 terabytes of data, the study revealed a total of almost 100 billion trees across the country.
“Stuff like that kind of blows my mind. First and foremost, you're like, ‘Wow, that's a lot of trees’. Second, it was the first time that we actually even asked the question and got an answer to, ‘Where are all the trees in the United States?’ Even though we know that trees are going to be an incredibly important component to meeting any kind of national or international climate goals.”
Since its launch, Microsoft has expanded Joppa’s baby into the fields of humanitarian action and accessibility. It feels like a no-brainer. What else should a company with so much computing power at its fingers do other than put that power at the hands of innovators who need it?
But it took someone like Joppa to make that link. The fact that he did speaks to his atypical way of seeing the world.
His aspiration, he says, is to 'always be the dumbest person in the room'
To an extent, this is innate. His aspiration, he says, is to “always be the dumbest person in the room”. For some, this may have the ring of false modesty (note: Joppa whizzed through a PhD in conservation and theoretical ecology in less than three years). Yet, in his case, it reflects an incurable intellectual curiosity.
It’s a quality the recruitment team at Microsoft spotted early. Unsure what to do after finishing his post-graduate studies at Duke University, an academic colleague pointed him towards a new venture within Microsoft Research Lab looking into computational ecology. The team was based in Cambridge, UK, and Joppa flew over for a gruelling, day-long interview.
“It was just 20 minutes at a time standing up at a whiteboard, one on one with some of the world's most brilliant computer scientists . . . I just remember wandering around Cambridge kind of in a daze thinking, ‘Well, that was embarrassing. I'm glad this was in England where nobody knows me’.”
It worked out better than he thought. What impressed them wasn’t the accuracy of his answers, he reflects. Instead, it was his honesty about the limitation of his knowledge and his “willingness to work through it” that stood out.
“At a place like Microsoft, they're looking for your curiosity. They're looking for your willingness to ask questions and to listen and to learn from other people.”
Along with his natural curiosity, Joppa’s instincts also push him towards this kind of empathetic and collaborative mode of working. Between his studies, for example, he spent two years working for the Department of National Parks and Wildlife in Malawi as a Peace Corps volunteer.
He credits the experience with not only reinforcing his lifelong passion for nature, but also for opening his eyes to the daily reality of on-the-ground conservation; namely, undertaking mammoth tasks with (all too often) poultry budgets.
I had my eyes just blown wide open by working every day with 2,000 of the world's leading computer scientists and seeing what they're capable of doing
Of course, AI for Earth isn’t entirely down to Joppa’s natural talents. He describes the five years he spent in Microsoft’s Cambridge research team as a pretty unique opportunity to live what closely resembled “an academic life embedded in this massive corporation”.
“I had my eyes just blown wide open by working every day with 2,000 of the world's leading computer scientists and seeing what they're capable of doing, [and] what Microsoft and its software was capable of doing.”
Joppa isn’t done yet either. Expect AI for Earth and its affiliate programmes to evolve and grow. Anticipate other similarly innovative ideas too. How can we be sure? Because Microsoft’s first ever global chief environment officer is not good at sitting still.
By any measure, Joppa is on a stellar career trajectory. Yet this 37-year-old over-achiever is not motivated by success for its own sake but by the platform that success offers to do something different, to push a new and unexpected door.
So at every juncture of his career Joppa has asked himself the same question: “What's the opportunity that's available to you because of something that you did that nobody else has an opportunity to do?”
This goes a long way to explaining his professional capacity for disruption. And, yes, it might just leave him as the dumbest person in the room. But not for long. Joppa is far too curious and keen to learn for that.
Lucas is speaking at this year's Responsible Business Summit West on the Environment Keynote: Role of technology & innovation in solving environmental challenges. Lucas will be joined by 200+ CEOs, heads of business and Investors sharing ideas on how business can deliver the circular, clean-tech economy. To see the full speaker line-up, click here
CV Lucas Joppa
Chief Environment Officer, Microsoft
July 2018 – present
Chief Environmental Scientist, Microsoft
March 2017 – July 2018
Computational Ecologist, Microsoft
July 2009 – February 2017
PhD in Ecology, Duke University
2006 – 2009
Peace Corps, Malawi
2004 – 2006
B.S. in Wildlife Ecology & Zoology, University of Wisconsin
2002 – 2004
This article is part of our monthly Disruptor series and also featured in our December magazine along with:AI for Earth iNaturalist Wild Me SilviaTerra AI for good