Posts

, , ,

Tech Giants are Using AI to Meet the SDGs

AI to Meet the Sustainable Development Goals
Tech giants are using artificial intelligence (AI) and machine learning (ML) to create innovative strategies to meet the United Nation’s Sustainable Development Goals and eradicate global poverty by 2030. A central barrier to development in third-world nations is in-access to high-quality, timely and accessible data.

Big data platforms like AI expand capabilities to acquire accurate, real-time, micro-level information, while ML allows pattern recognition at a macro-level. Combined, these data advances can make data more accessible, applicable and finely scalable while accelerating the speed and scale for private and public development actors to implement change. Companies are partnering across public, private and nonprofit sectors to broaden the collective impact.

Take a look at the innovative approaches tech giants are taking to help global poor communities with data and what the incorporation of AI technologies means for the future of global poverty initiatives. These approaches aim to employ AI to meet the SDGs within its allotted time frame.

Education and Digital Training

On June 19, 2019, the day preceding World Refugee Day, Microsoft announced the inception of two projects partnering with Asylum Seeker Advocacy Project (ASAP) and Kids in Need of Defense (KIND). These projects supplement its AI for Humanitarian Action group to help incorporate AI to meet the SDGs.

The AI for Humanitarian Action group is a $40 million, five-year program part of Microsoft’s larger AI for Good suite (a $115 million, five-year project). The projects will provide AI tools to help staff track court dates, prioritize emergency cases and translate for families with AI speech-to-text. Microsoft also has continuing partnerships to incorporate AI/ML into educational services for refugees with the following groups:

  1. International Rescue Committee (IRC): This committee works to provide humanitarian aid through the creation of sustainable programming for refugees, displaced populations and crisis-affected communities. This includes career development programming and digital skills training to empower refugees and make them relevant for the job markets in each affected country. Microsoft and IRC’s Technology for Livelihoods in Crisis project in Jordan is an example of this.
  2. United Nations Children’s Fund (UNICEF): Together with the University of Cambridge, the UNICEF is developing The Learning Passport. The digital platform will ensure better access to education and facilitate learning opportunities for youth displaced by conflict and natural disasters. It creates scalable learning solutions tailored to each child. Crises have affected the quality of education for 75 million youth.
  3. Norwegian Refugee Council: This council is providing an AI chatbot service that uses language understanding, machine translation and language recognition to deliver high-quality education and digital skills training to refugees. This helps to close the education gap for the millions of youth affected by conflict. It will also help humanitarian workers communicate with migrants who speak other languages, which will help them best provide the best service.
  4. United Nations High Commissioner for Refugees (UNHCR): UNHCR plans to provide 25,000 refugees in Kakuma with access to high-quality, accredited, context-appropriate digital learning and training by 2020 for development in Kakuma markets. UNHCR intends this project to expand across multiple countries.

Food Security and Agricultural Development

The fact that farms do not always have power or internet security limits technological developments that address food security and agricultural development. Here are some efforts that consider the capabilities of farmers and the respective developing regions:

  1. Microsoft FarmBeats: It aims to enable data-driven farming compatible with both the capabilities of the farmer and the region. FarmBeats is employing AI and IoT (Information of Things) solutions using low-cost sensors, drones and vision and ML algorithms. This combined AI and IoT approach enables data-driven improvements in agriculture yield, lowered costs and reduced environmental impacts of agricultural production, and is a significant contribution to help AI to meet the SDGs.
  2. Apollo: Apollo uses agronomic machine learning, remote sensing and mobile phones to help farmers maximize profits in developing markets. Apollo delivers scalable financing, farm products and customized advice to farmers while assessing the farmers’ credit risk. Apollo customizes each product in order to double farm yields and improve credit. The beta project is starting in Kenya.
  3. The International Center for Tropical Agriculture (CIAT)/CGIAR research group: It aims to implement preemptive solutions rather than reactive solutions to end hunger and malnutrition by 2030. CIAT has developed a Nutrition Early Warning System (NEWS), which uses machine learning to make predictions on malnutrition patterns based on current and future estimates of crop failures, droughts and rising food prices. This approach is able to detect an impending nutrition crisis and take action instead of responding after the crisis has taken hold.

Socioeconomic Data Collection

According to a report by The Brookings Institute as a part of its “A Blueprint for the Future of AI series,” data providing national averages “conceal more than they reveal” and inaccurately estimate and map patterns of poverty. Survey data is often entirely unavailable or otherwise low in quality in many of the poorest countries where development needs are greatest. 39 of the 59 countries in Africa conducted less than two surveys between 2000 and 2010.

Even in large countries with sophisticated statistical systems, such as India, survey results remain inaccurate, with the gap between personal reporting and national accounts amounting to as much as a 60 percent difference in some countries. Companies are addressing this by utilizing big data from remote sensing satellites.

The Group on Earth Observations (GEO) is using Earth Observations (EO) to provide finely-tuned and near-real-time data on economic activity and population distribution by measuring nighttime luminosity. Researchers have noted a correlation between luminosity and GDP as well as subnational economic output. Collecting socioeconomic data in this way can ensure higher quality data important to policy implementation and direction to countries with the greatest development needs.

Timothy Burke and Stan Larimer launched Sovereign Sky in 2018, putting satellite data into action. Sovereign Sky is the world’s first space-based blockchain which provides secure private internet networks and powers a new Free World Currency to redistribute the world’s wealth with a goal of eradicating poverty by 2032.

The eight current satellites cover Africa and India and the organization will send boxes of StealthCrypto phones, digital wallets, smart cards and modems to people in need. Sovereign Sky will deploy 36 satellites within three to 10 years to cover the entire world in a secure blockchain internet connection, closing the gap on technological interactions between all nations and including the world’s remotest and poorest areas in internet connectivity.

Pitfalls of AI-Driven Global Development Initiatives, and Moving Forward

AI and ML have crucial capabilities in reshaping education, agriculture and data collection in the developing world. However, these technologies have a history of producing unethical racial profiling, surveillance and perpetuating stereotypes, especially in areas with a history of ethnic conflict or inequality. AI and ML applications have to adapt in ways to ensure effective, inclusive and fair distribution of big data resources in the developing world. Development experts need to be in close collaboration with technologists to prevent unethical allocations.

This diversification is why it is important that tech giants like Microsoft, and projects like those by the ICAT/CGIAR, are created in collaboration with various nonprofit, public and private sector groups to ensure interdisciplinary ethical liability for big data applications in sustainable development contexts. Ensuring the use of AI technologies is context-specific to the affected regions and populations will help prevent misappropriation of the technology and increase quality and effectiveness.

Working with local companies and sectors can create long-lasting engagement and grow permanent technology sectors in the developing areas thus contributing to the local economy. These strategies can put forth effective, ethical and productive applications of AI to meet the SDGs.

– Julia Kemner
Photo: Flickr

, , , , ,

A Look at the Positive Side of AI and Drones

Drone
No technology is inherently good or bad; rather, it is humanity’s use of that technology that can be evil or virtuous.  Certain modern tools seem only capable of carrying out despicable or ultimately evil deeds as controversy surrounds them, and their names evoke fear. Artificial intelligence (AI) and drones are two of the most widely commented on and feared applications of modern science. Despite the prevailing negative perceptions, AI and drones are also used for a good cause: combatting poverty.

Unequal Scenes

Although drones, or UAVs (unmanned aerial vehicles), are often used in violent attacks and warfare, they and their human operators are doing wonderful things across the world. Photographer Jonny Miller used drones to capture cityscapes and the line dividing the rich and the poor. He captured images of lush, green golf courses directly up against dirt roads and shack neighborhoods. Giant mansions can be seen with trees and acres of grass next door to brown areas with buildings packed into a small plot. Miller’s project “Unequal Scenes” is raising awareness about poverty and inequality which would be impossible without drone photography.

The Problem of Land Ownership

More than half of the world’s population, usually women, cannot prove that they own their land. This is especially problematic in the country of Kosovo, where most of the men and boys were murdered during the Balkan wars of the 1990s. The women who remained have worked tirelessly to rebuild their homes and communities, but they face an enormous roadblock: the inability to use their vast land resources to provide for themselves economically. These women do not have any sort of documentation for their lands once owned by their husbands. One woman explained that she had applied for loans to build her business but was repeatedly turned down because she lacked what the government called “property documents to put down as a guarantee.”

These communities do not have the means to hire land surveyors necessary for official registration. Property owners with potentially good, profitable land are powerless without official documentation. However, drones are helping these women. The World Bank Group’s Global Land and Geospatial unit dispatches drones to map out land plots. Drones survey and map for a fraction of the cost of traditional means, giving the Kosovan women the ability to register their lands and ultimately invest in their own property.

The Positive Impacts of AI

Artificial intelligence (AI, also referred to as “machine learning”) refers to a machine’s ability to imitate intelligent human behavior. AI is often associated with 1980s movies about robots destroying humanity based on a real fear that one day the machines will become self-aware and grow tired of serving humanity; “the development of full artificial intelligence could spell the end of the human race,” warned Stephen Hawking in 2014. Despite this apparent destructive potential of AI, it is currently transforming agriculture and changing the African business environment in the real world.

One writer argues that Africa is amid the “fourth industrial revolution … ushered in by the power of AI.” Many innovative African business leaders have embraced AI to improve productivity and efficiency. One example is a Moroccan company which uses AI to perform analytics on data sent from devices on motorcycle helmets. This improves riding habits and provides more accurate insurance premiums, reducing costs and improving safety for riders. Another instance involves an Egyptian manufacturer using AI to automate certain processes and reduce overall error while improving quality of service, which ultimately reduces the cost to the consumer. Finally, one Algerian firm helps local doctors provide cancer detection and treatment for their patients. The firm uses AI to create models that can diagnose those who are unable to visit hospitals for formal examinations. This has the potential to save the lives of many who don’t have the means to get regular checkups and screenings.

In addition to previous models, AI is also reducing overall costs for farmers and helping to improve their yields in India. Certain Indian dairy cows are given radio-frequency identification tags that transmit important information about the cows’ diets and overall health to cloud storage where it is “AI-analyzed.” The farmers receive alerts about any potential issues of the cows that require their attention. This can reduce costs and increase efficiency for the farmers.

These are just some of the ways that technology often labeled as “bad” is being used for good, especially in the fight against poverty. Cases like these prove that technology cannot be inherently evil and that there are good uses for AI and drones. While some individuals use modern equipment to destroy the world, there are plenty of men and women using the same tools to improve it.

– Sarah Stanley

Photo: Flickr

, ,

The Age of Machine Learning: Eradicating Poverty

Machine Learning
In today’s information age, the most abundant resource has quickly become information itself, more specifically data. By 2010, the world had created 1.2 zettabytes (1.3 trillion gigabytes), an equivalent to 75 million 16 GB iPads. By 2016, the world completely surpassed this, creating 90 percent of all data in just the last two years. 2.5 quintillion bytes of data are created every day.

These numbers are much too large for any single person to comprehend, but with the help of technology and machine learning the data can help optimize transportation in cities, predict the stock market and diagnose diseases, along with a vast amount of other tasks. Big data has largely been a tool for the developed world; however, there is plenty of potential for it to become a key factor in ending poverty.

In August, Stanford researchers published a paper on using satellite imaging and machine learning to track and measure poverty throughout Africa. Accurate measurement of poverty in Africa was extremely lacking as “39 out of 59 African countries conducted fewer than two surveys to measure poverty” between 2000 and 2010, according to The World Bank.

Previous strategies for measuring poverty also included tracking mobile phone usage and satellite photos of lights at night; however, phone data was not always available and nighttime light data could not differentiate between poverty and extreme poverty levels.

Instead, the Stanford researchers used daytime images of development, such as paved roads, farms, metal roofs, along with nighttime light intensity data to measure poverty. They input the data into a computer model to map out poverty levels throughout the test countries, Nigeria, Uganda, Tanzania, Malawi and Rwanda. Using methods such as these, African governments and NGOs will be better equipped to design policy and find areas most affected by poverty.

Furthermore, one of the most famous machine learning tools is IBM Watson, a supercomputer that uses advanced software to answer questions. In 2014, IBM launched Project Lucy, a mission to bring Watson to Africa and use the artificial intelligence to help solve the problems surrounding “health care, education, water and sanitation, human mobility and agriculture.”

More generally, scientists predict that machine learning has the potential of predicting the future and keeping watch over society. More specifically, the technology has the capability of forecasting underperforming crops in developing countries and situations that will call for an international convention.

Using biometric data, governments, especially that of India, hope to identify all citizens and ensure they can receive subsidies and benefits, helping to end inefficiency and corruption.

Machine learning is clearly a revolutionary technology, but its true potential is still unclear. So far, it acts as an aid to researchers, aggregating data and producing summaries.

However, machine learning could even advance to levels of innovating on its own. For example, instead of diagnosing a disease, machine learning could help find the cure to one. In the next decade or so, the world will wait and see where this amazing technology can take it.

– Henry Gao

Photo: Flickr