How Microsoft and Amazon are making AI more real
Adapted from the VERGE Weekly newsletter, published Wednesdays.
Species identification schemes. Early warning systems informed by sensors and big data. Predictive population modeling. Looks like a job for artificial intelligence!
It felt appropriate to weigh in with a year-end riff about a topic that’s guaranteed to spark debate at holiday parties — when and how to use AI for jobs that humans just can’t perform as efficiently as smart software.
This somewhat hypothetical question has special relevance in the context of recent developments at Amazon and Microsoft, two tech giants seeking to outwit each other in establishing their cloud software services as platforms for running and managing AI applications.
On Dec. 10, Amazon Web Services quietly launched what it’s calling the Amazon Sustainability Data Initiative, which builds on the "vast amounts of data that describe our planet." The idea is to get researchers to store oodles of weather observations and forecasts, satellite images and metrics about oceans, air quality — you name it — so that they can be used for modeling and analysis. And then, it encourages organizations to use the data to make decisions that encourage sustainable development.
One example of how the data is already being used: the Famine Action Mechanism, developed by the United Nations, World Bank and International Committee of the Red Cross. The idea is to monitor indicators about the diverse causes of food insecurity such as droughts, flooding, regional conflict, food pricing and economic policies so that regions of concern can be identified proactively — and so that funding and resources can be applied to preventive full-fledged crises. Naturally, this resource is being hosted on Amazon cloud services.
Microsoft is also busy: it just disclosed 11 more research projects that will receive a total of $1.28 million in new grants under its $50 million AI for Earth initiative, its bid to help scientists, startups and academics apply artificial intelligence to complex research aimed at addressing — or at least better understanding — climate change. One thing the Microsoft team is really prioritizing is how AI can help preserve biodiversity.
"There are species disappearing off our planet that we’ve never even known about," notes Microsoft senior director Josh Henretig, pointing to one project AI for Earth is funding, run by an outfit called iNaturalist. "We also don’t understand the reasons or implications for why and how quickly they may be disappearing. We don’t have the ability to scale up millions of field researchers — but we do have the ability to tap millions of smartphone users to help collect that data quickly."
The latest AI for Earth recipients were selected by Microsoft and National Geographic in tandem — and they’re pretty similar to some of the other research initiatives that the program has supported in the past year: Microsoft’s machine learning algorithms will aid in making decisions about irrigation development and crop water efficiency, mapping dams and reservoirs and monitoring ecosystem health by classifying the songs of insects in tropical rainforests.
The grants include financial support, access to Microsoft’s cloud and AI tools and an affiliation with National Geographic Labs. "Human ingenuity, especially when paired with the speed, power and scale that AI brings, is our best bet for crafting a better future for our planet and everyone on it," said Lucas Joppa, chief environmental officer at Microsoft, in the press release describing the latest awards.
Indeed, there’s more evidence than ever that AI is a force that everyone involved in sustainability should continue to watch closely, as we suggested earlier this year in the State of Green Business report for 2018. Google’s experiments in using AI for data center management deserve continued scrutiny, as do intriguing startups such as Smarter Sorting, which has raised $9.3 million in seed funding for an AI service that can be used by retailers to automate waste management decisions for returned goods.
Incidentally, Microsoft and Amazon are onto something. One of Deloitte’s predictions for 2019 is that most businesses — up to 70 percent — eventually will turn to cloud-based AI services to take advantage of these capabilities. By 2020, the consulting firm predicts, up to 87 percent of enterprise software apps will have AI built in. That sounds pretty real to me.
And for now, at least, the sorts of projects being funded by the likes of Amazon and Microsoft have clear benefits that outweigh many legitimate concerns frequently raised about AI, such as the potential for ethics breaches, for gender and ethnic discrimination in decision-making and for erasing jobs currently held by people.