Co-authored with Mohsen Gul (University of Nottingham). This is based on research conducted for UNDP Innovation Unit Africa.

For the purpose of this article, the Global South refers to countries in Africa, Latin America, and developing Asia including the Middle East.

1 Current Landscape

To date almost all of research work has been focused on the implications of frontier technologies like Artificial Intelligence (AI), Machine Learning, automation and Internet of Things (IoT) for people living in higher income countries such as the EU, UK and US. The focus of this article is to identify key examples on the impact of artificial intelligence as a frontier technology on citizen engagement in low and middle-income countries for whom many of the same opportunities and risks apply as in the higher-income courtiers (often to a greater extent), along with additional opportunities and risks unique to these countries.

According to PWC (2017), China stands to see the biggest economic gains with AI enhancing GDP by 14.5% in the coming years but it notes that most of the other developing countries will experience more modest increases due to the much lower rates of adoption of AI technologies expected. China’s State Council has recently issued guidelines on AI development wherein it is aiming to become a global innovation centre in the field by 2030, with an estimated total output value of the AI industry projected at $147 billion. China Artificial Intelligence Industry Innovation Alliance (CAIIIA) was set up in 2017. The newly formed alliance set a target to incubate 50 AI-enabled products and 40 firms, launch 20 pilot projects, and set up a technology platform in the next three years. Singapore recently announced plans to invest over $100 million in AI over the next five years. Measured by patents filed, from 2010–14, China was second submitting 8,410 AI applications. During this period, Japan and Republic of Korea submitted 2,071 and 1,533 respectively. India was also among the top 10 countries globally in terms of numbers of patents submitted. In addition, China and India are among the top 10 countries in terms of the number of AI companies (ESCAP, 2017).

There are plenty of instances where AI is being used to improve delivery of public services and public goods in low and middle-income countries, ranging from pilot projects to larger scale roll out. AI seems particularly fit for simplifying transactions on government websites. AI has also been deployed as a response to public health concerns, such as to anticipate outbreaks of diseases such as Zika and dengue fever. For example, the Brazilian NGO Viva Rio partnered with start-up AIME (Artificial Intelligence in Medical Epidemiology) which analyses existing local government datasets in combination with satellite image recognition systems to deploy low cost predictions of where we should expect greater incidence of disease in an upcoming three-month period. Notably, the technology was designed to work in Malaysia, but it had success rates of 84% diagnosis in Brazilian trials. Following its low-cost success in Brazil, AIME has been deployed in Dominican Republic.

In other cases, AI has been used to improve police coverage, such as in dealing with transit issues. In Uganda, AI is used to advise individuals or emergency vehicles on optimal routes, dynamically redeploying a limited number of traffic police officials, and analysing possible reconfigurations of the road network to remove bottlenecks. In other cases, it has been used for environmental ends. In Kenya, for example, the World Wide Fund for Nature (WWF) supports the use of an AI device with drones. After nine months, over a dozen hunters had been apprehended in the Maasai Mara. WWF has received a $5 million grant from Google.org to employ this AI-powered device to protect wildlife. AI has also been used for agricultural matters, including identifying crop disease with a smartphone. Mcrops, developed in Uganda, is a diagnostic tool for diagnosing viral crop diseases in cassava plants. Sick plants are flagged in real time, which allows farmers to take action and stop the spread of the disease. In Nigeria, AI is being employed to help farmers sell their produce and buy services via a bot platform that relies on SMS and other channels such as USSD, Slack, etc.

AI has also been used to prevent and predict natural disasters. The Red Cross/Red Crescent Climate Centre has an on-going project with Togo’s Nangbéto Dam, which frequently overspills, causing great disruption to the livelihoods of people living downstream. In the past models were poor at predicting the likelihood of overspill but using a combination of crowdsourced information (including by mobile phone) and AI techniques, an improved model of overspill prediction was developed. Also, the Netherlands Red Cross’ data initiative has developed models using data from previous natural catastrophes to allow better resource management and prioritization of aid. It was employed after Typhoon Haima in the Philippines in 2016 and proven to be accurate after a comparison with data on damaged property.

AI-based automated translation and voice recognition systems are also emerging which can have significant impact in countries with multiple languages. This is the case for numerous low and middle-income countries, including India, Indonesia and Nigeria. These systems could also have an impact in places with high levels of illiteracy, allowing people to engage with the government or public service provision interfaces by spoken rather than by written means. This would particularly benefit marginalised groups who experience disproportionate rates of illiteracy. Some of the examples of automated translation and voice recognition systems include ‘Translator Gator’ developed by Pulse Lab Jakarta (Indonesia) to invites people to create taxonomies, or collections of keywords, in lesser-known languages and dialects, and ‘Conversations as a Platform’ developed by Microsoft and Government of Singapore to represent a fundamental shift where the power of human language will be used to make chatbots more anticipatory, accessible, engaging and inclusive for citizens and constituents.

Some case studies are discussed in detail in the next section. For each case study, information about geographical coverage and scale of operation is provided.

1.1 Case Studies

1.1.1 Asia

AI for Parliament in Japan (Japan, Asia; Scale: Pilot)

The Japanese Government is trialing artificial intelligence to help officials draft responses submitted to the parliament. The tech would help officials draft responses used in policy-making, by mining past opinions on policy issues and alternative suggestions voiced out by officials during parliament. The government will first feed five years’ worth of parliament agenda summaries to the system before it churns out responses.

Based on trial results, Tokyo will then expand the use of AI in other branches of government. The government has launched a broad plan to use AI and robotics across public services. Called “Society 5.0”, the government envisions using science and technology [to] play even more key roles in tackling challenges like ageing, sluggish productivity growth and enhancing [the] wellbeing of humans.

Singapore’s Dengue Risk Map (Singapore, Asia; Scale: Operational)

As a tropical country, Singapore faces seasonal bouts of dengue cases, peaking in the warmer months of June to October. The country’s main strategy is to carry out preventive surveillance of dengue hotspots. However, this is labour-intensive, and drains an already limited pool of skilled vector control officers. So, the Government decided to use data to identify the areas most at risk of dengue — and help the agency deploy officers more effectively.

The team used a machine-learning model to create a colour-coded risk map. The model analyses data from various agencies to rank the overall risk of dengue across the entire country. These data include past dengue cases from the Ministry of Health; population estimates from Singstat; vegetation indices from the Centre for Remote Imaging Sensing & Processing at the National University of Singapore; public transport information from Future City Lab ETH-NUS; and mosquito population data from NEA’s surveillance programme.

The risk map also enables targeted publicity campaigns — for example, placing advertisements about dengue prevention at bus stops in high-risk towns. The project won Best Use of Data at the GovInsider Innovation Awards, held at Innovation Labs World in 2017.

mAadhaar app (India, Asia; Scale: Operational)

The Unique Identification Authority of India (UIDAI) has launched the mAadhaar app, which allows users to carry digital versions of their identification profiles on their phones. Users of the app may download their Aadhaar number profiles on their smartphones and will therefore not need to carry the physical cards with them, The Economic Times reported. The Aadhaar number is a 12-digit random identification number issued by the UIDAI to Indian citizens, which is linked to basic demographic information such as name and date of birth, and biometric information such as fingerprints and iris scans. These data are stored in a centralised biometric ID database. The Aadhaar system has over 1.1 billion citizens enrolled, and is saving current administration about US$2 billion a year.

In November 2017, The Ministry of Commerce and Industry in India has set up a Task force on Artificial Intelligence to kick-start the use of AI for India’s economic transformation.

SAAD (Dubai) (United Arab Emirates, Asia; Scale: Operational)

The city’s first government service utilising artificial intelligence, powered by IBM Watson. It was aimed at allowing entrepreneurs and investors to ask questions related to setting up a business in Dubai, and to get real-time responses on various topics, including business licensing requirements and registration processes. “Saad” is designed to understand natural language and ingest and comprehend massive amounts of data, learn and reason from its interactions, and provide responses that will aid users in deciding on correct courses of action.

VAMPIRE tool (Indonesia, Asia; Scale: Operational)

Indonesia is under threat of food insecurity and undernutrition, and faces “alarmingly high” stunting levels. Pulse Lab Jakarta joined forces with the World Food Programme (WFP) and the Food and Agricultural Organisation to develop a platform that identifies affected communities in need of help. The VAMPIRE visualisation tool combines data in several layers: users can view where food-insecure and agriculture-dependant communities live; data on rainfall anomalies and vegetation health; and crowdsourced prices of staple foods in these areas. VAMPIRE, which is installed in the President’s situation room, is “an early warning system” for authorities to manage food security and better allocate resources. Following VAMPIRE’s success in Indonesia, WFP is now in discussions with the Sri Lankan government to have it installed there too.

Bindez (Myanmar, Asia; Scale: Operational)

Bindez emerged out of a challenge that has faced many countries coming online in the last few decades: how to make their countries languages talk with computers. Bindez is an acronym for Burmese Index, a project that started as a local language search engine by 2 Burmese techies Yewint Ko and Htet Will in mid-2013, shortly after the country had opened up to new technology, and content consumption needs became evident. It now uses natural language processing and text analysis to track online hate speech.

1.1.2 Africa

Sustainability and Artificial Intelligence Lab (Africa; Scale: Operational)

The Lab is combining satellite imagery and machine learning to predict poverty. Using the final model that has been trained on survey data, it can estimate per capita consumption expenditure for any location where we have daytime satellite imagery. The Lab is commissioned by Stanford’s Global Development and Poverty Initiative and partially supported by the National Science Foundation.

Robotic Pharmacists for HIV Patients (South Africa, Africa; Scale: Pilot)

Robotic pharmacists are dispensing drugs to people with HIV on the streets of South Africa. The country has the highest number of people living with HIV — 7 million. However, 3.3 million people are not on drugs they should be, according to South Africa’s Department of Health. In the past, the antiretroviral drugs have been available only via expensive and private care.

The robots will not reveal patient identity helping to remove any social stigma associated with the disease. It will also dispense medicines also for other patients with chronic diseases like TB. With the robot dispenser, patients may no longer need to wait for hours at hospitals or clinics to get their monthly dose. The pilot is run by the Right to Care project in Helen Joseph Hospital, Johannesburg funded by the Department of Health and non-profit The Global Fund.

Similar Initiative: Nigerian startup Aajoh uses artificial intelligence to help individuals that send a list of their symptoms via text, audio and photographs, to diagnose their medical condition.

1.1.3 Central and South America

Rosie (Brazil, S. America; Scale: Operational)

In Brazil, a group of data-analysis experts has used artificial intelligence techniques to monitor public officials. They focused on limiting fraud among members of congress seeking reimbursements for their travel and food expenses; after getting crowdfunding for the startup costs, they created Rosie, an Artificial Intelligence robot that analyzes the reimbursement requests of lawmakers and calculates the probability that they are justified. To no one’s surprise, Rosie found that the deputies often cheated.

Similar Initiative: The Serious Fraud Office in the UK developed a robot with help of AI startup RAVN which then helped the government’s law enforcement arm sift, index and summarise 30 million documents relating to fraud investigations.

CitymisVis (Argentina, S. America; Scale: Operational)

An interactive tool developed by Government of Argentina in collaboration with Microsoft Latin America that allows citizens to report and track problems from poor street infrastructure to deficient urban sanitation. Through this tool, municipalities listen and respond to requests and complaints from the citizens. As a result of the successful adoption of Citymis Community in several municipalities, large amounts of data are daily generated, which require analysis by the corresponding service departments. The intelligent analysis of citizen requests and complaints can lead to improved levels of service coordination and can help in the decision-making process by relevant authorities.

Minecraft (Mexico, C. America; Scale: Operational)

A United Nations initiative is using the computer game Minecraft to help citizens design public spaces in more than 25 developing countries including Mexico. Called Block by Block, the UN project turned the game into a ‘community participation tool’ in urban design, with a focus on poor communities. Minecraft is the world’s second best-selling videogame of all time. In the game, players use textured cubes to build a virtual world. According to Microsoft Research’s Cambridge lab, people build amazing structures that do amazing things in Minecraft, and this allows experimenters to put in tasks that will stretch AI technology beyond its current capacity.

Elefantes Blancos App (Columbia, C. America; Scale: Operational

$163 million in public works corruption and inefficiencies has been identified citizens using the smartphone app, Elefantes Blancos (White Elephants). The smartphone app was introduced in 2013 by Colombia’s Transparency Secretariat and since that time 54 “white elephant” projects — those projects perceived by citizens as public works that were languishing, unfinished or abandoned — have been identified. 15 of the projects worth more than $400,000 were completed with pressure from the Transparency Secretariat, 27 are under prosecution and 12 projects are currently being investigated (Guay, 2017).

Citizens can download the app to their smartphone and then upload photos of the unfinished public projects. Users can vote on the most problematic projects, and the app determines location and frequency of reporting for each project. The Transparency Secretariat then begins the process of assessing the projects for corruption. While the Transparency Secretariat cannot prosecute cases, the Elefantes Blancos app has proven its value, and control of the app will be transferred to the local comptroller — the comptroller has the legal authority to prosecute corruption.

1.1.4 Challenges

It is important to note that countries in this geographical region are diverse politically, economically and socially. Hence, AI technologies pose several challenges at national and regional levels. Some of the key impediments identified include:

• A possible frontier technology divide emerging with advanced economies like Singapore and Indonesia investing in technological leapfrogging at astonishing rates, leaving others in the region at a risk of being left behind
• Imbued trust deficit amongst countries, private and development sectors exists regarding use and abuse of such technologies and associated data
• Limited technical capacity of governments and policymakers in developing technologically advanced governance solutions is leading to overdependence on donor agencies and other developed countries
• Limited research and baseline data are available to prioritise infrastructural elements of national and regional decision support systems
• A potential concern about technological unemployment persists due to surge in technological adaptation.

The article acknowledges that some countries in the Global South are leading from the front and are forecast to be the major market of the future. This prominent position means it is prudent for governments to think carefully about the policy priorities and issues raised in this article to address ethical issues, ensure development of an adaptable workforce for the future, and put in place regional level systems to allow sustainable technological innovation.