When the Google AI Impact Challenge, an open invite to organizations around the globe to submit their ideas for how AI could tackle social, humanitarian and environmental challenges, was launched in October 2018 it received 2,602 applications from six continents and 119 countries. After review by a panel of experts that included Jacquelline Fuller, Vice President Google and President of Google.org, 20 proposals from Australia to Uganda were selected for the program. As grantees, each received access to a support package that included coaching from Google’s AI experts and a US$25 million pool of Google.org grant funding.
With AI and its subsets, machine learning and deep learning, now infiltrating our lives on a daily basis, from asking Amazon’s Alexa for the morning news to snapping a photo on your smartphone and uploading it to social media, it’s little surprise that individuals and organizations worldwide are exploring how we can harness this technology to address problems of significant social importance. And, while many initiatives are still in the concept phase, analysts such as the Berkeley, California-headquartered not-for-profit AI for Good Foundation suggest that AI is already having a positive impact across all 17 of the UN Sustainable Development Goals.
In its ‘Applying AI for Social Good’ discussion paper, McKinsey Global Institute pinpointed 18 AI capabilities that can be applied to social issues, the majority of which (14) fell into three primary categories: computer vision, natural-language processing, and speech and audio processing. When these capabilities were mapped across 10 social domains (such as education, environment, and health and hunger), image classification and object detection stood out for their potential impact across a large number of social causes.
“These are still the early days of AI’s deployment for social good, and considerable progress will be needed before the vast potential becomes a reality.”
There are some important barriers to the successful implementation of AI for good solutions, however. Along with data accessibility and a lack of qualified AI professionals, one of the biggest roadblocks is in-house expertise: organizations either possess social sector or AI technical know-how, rarely both. Which is where initiatives such as Google’s AI Impact Challenge or enterprise AI platform DataRobot’s AI for Good program find their place to foster partnerships that turn ideas into impactful solutions that become user-friendly and self-supporting in the long run.
Although both Google and McKinsey caution that AI “is not a silver bullet”, the influence AI-based algorithms can have on some of the world’s most challenging issues are already in evidence, from using image recognition to identify disease in cassava, a plant that feeds more than half a billion people in Africa daily to using drones to monitor the endangered sea cow in Australia. “These are still the early days of AI’s deployment for social good, and considerable progress will be needed before the vast potential becomes a reality,” the McKinsey discussion paper states.
Here are three initiatives that demonstrate the scope of impact that can be created when using AI for social good.
Air pollution in the Ugandan capital, Kampala, is six times higher than World Health Organization guidelines, but researchers at the country’s Makerere University have their sights set on changing that with its AirQo prototype.
One of the 20 grantees of the Google AI Impact Challenge, the project has involved the placement of low-cost air sensors on motorcycle taxis. By applying AI to the sensor data, the researchers are able to not only monitor current air quality but also use the information to forecast more reliably.
“The data collected is used to inform public policies on reducing, containing and better management of air pollution and its associated health risks. The same data is also used to raise awareness on air quality issues,” said Dr Engineer Bainomugisha, Project Leader and Chair, Department of Computer Science, College of Computing and Information Sciences, Makerere University.
According to the World Bank, approximately 1.7 billion people are “unbanked” – or unable to access the services offered by a bank or banking institution. Founded in 2005, not-for-profit Kiva offers crowdfunded loans in its mission to help those very people have access to the financial services they need, whether that’s to start a business or further their education.
To ensure visibility for every loan request, no matter the geographical location of the applicant, the San Francisco-based application partnered with the DataRobot AI for Good program. Drawing upon data about each loan, a machine-learning solution has been built and integrated into Kiva’s website to ensure no application slips through the cracks.
“Our goal was to get loans in front of people that might not otherwise be seen, and we can do that with machine learning,’ says Brian Kimming, Kiva’s Senior Data Scientist.
Another of the Google AI Impact Challenge recipients, the brains behind SkillLab have one mission: “No person should be excluded from opportunity simply because their skills are invisible”. The Amsterdam-headquartered organization champions an inclusive labor market where opportunities for social and economic participation are open to everyone, no matter their status, race, gender or background.
SkillLab’s AI-powered assessment mobile app allows refugees and migrants in particular to document their skills in their native language and, in turn, be guided towards relevant career opportunities and education pathways in local labor markets.
“The vision behind the tool is that employment services can use it in their first contact with newly arrived refugees, and build them a great profile on which personal career planning services can be based,” SkillLab Co-Founder and Managing Director Ulrich Scharf said in an interview with the European Commission Electronic Platform for Adult Learning in Europe.