Can AI help tackle the most pressing challenges in developing countries?

This year, the GSMA Mobile for Development team is embarking on the journey to explore the opportunities and challenges of frontier technologies in the developing countries.

Through desk-based and field research, we will explore the applications of these technologies in the fields of education, health, agriculture, utilities access and more.

This blog is the first in the series which will focus on the role of artificial intelligence (AI) in addressing the UN Sustainable Development Goals (SDGs). The blog will highlight some promising use cases, as well as challenges in deploying and scaling AI solutions, in developing countries.

Artificial intelligence (AI) is the ability of a machine or computer to emulate human tasks through learning and automation.

These human capabilities are augmented by the ability to learn from experience and adapt over time. AI enables machines to sense their environment, think and learn to take action in response to specific environment and underpinning circumstances. Its applications can be narrow whilst only addressing a specific set of tasks with human-like ability e.g. language translation, self-driving cars etc. and, wider applications where the system can teach itself to interact with a broad variety of independent and unrelated tasks.

While the old perception about AI was that of a ‘western phenomenon’ with robots and ‘Jarvis-like assistants’, AI in fact has immense potential to have a sizable and pronounced impact on the UN Sustainable Development Goals (SDGs).

A few examples to illustrate this:

  • AI-powered analytics of crop data can help identify diseases and enable soil health monitoring (SDG 2)
  • AI can streamline the triage process, as well as improve diagnosis and post-care follow up procedures (SDG 3)
  • AI can drive balanced hiring recruitment and hiring practices and spotlight gender inequity (SDG 5)
  • Remotely collected metering data, when fed into AI, can predict consumption patterns for safe water provision (SDG 6)
  • Machine learning algorithms can improve photovoltaic energy capture, lowering the cost of solar power (SDG 7)
  • Data-driven and AI-powered urban planning can make cities smart and sustainable (SDG 11)
  • Computer vision technology can be trained to sort plastic waste (SDG 12)
  • AI can be used to predict climate-related disasters (SDG 13)
  • IoT sensors powered with AI can track the movement of fishes, detect the appetite of fish, and provide optimal fish feed, avoiding overfeeding or underfeeding (SDG 14)
  • Pattern recognition and wide applications of computer science can help combat illegal poaching (SDG 15)
  • Thoughtful integration of AI can help reduce corruption and discrimination (SDG 16)

Artificial Intelligence in Action

The examples listed above may well seem theoretical but a growing number of AI initiatives and solutions are already being trialled across developing markets, many that will create lasting socio-economic impact.

  • Optimetriks, a GSMA Ecosystem Accelerator Innovation Fund grantee (Uganda), leverages its network of crowdsourced field agents to provide regular data feeds on the needs, activities and stock levels of FMCG companies and mobile operators, helping these corporates to better understand their distribution channels. Optimetriks processes the collected data through real-time visualisation dashboards, and uses AI-driven image recognition to detect products and merchandising material. This raw data can be used to predict optimal stock-up levels.
  • Qlue (Indonesia), another GSMA Ecosystem Accelerator Innovation Fund grantee, runs a mobile-enabled platform that enables residents to request government services. City governments can respond and provide updates on the status of those requests. Qlue addresses cities’ safety and security issues in real time through mobile apps, AI-powered computer vision and GIS Integration dashboard.
  • The Child Growth Monitor platform by Deutsche Welthungerhilfe, aims to rapidly and accurately measure child malnutrition using augmented reality and AI technology to undertake the scans.
  • Refunite, a GSMA Disaster Response Innovation Fund grantee (Uganda), is a micro-tasking platform for refugees which provides them with a small income for each verified task they complete, using image categorisation for training algorithms.
  • Apollo Agriculture (Kenya) uses satellite imaging and agronomic machine learning to deliver agricultural input loans and customised agronomic advice to smallholder farmers via mobile phone.
  • Gringgo (Indonesia) uses an image recognition tool to improve plastic recycling rates, reduce ocean plastic pollution and strengthen waste management in under-resourced communities.
  • Niramai (India) provides a cost-efficient way to detect breast cancer and is a better alternative to the existing method of mammography which requires a high capital cost. The core of Niramai’s solution is Thermalytix, a computer-aided diagnostic engine that is powered by AI.
  • Ubenwa (Nigeria) developed an AI app that analyses a baby’s cry to give warning signs of asphyxia, which is the third leading killer of infants worldwide.
  • Saphron (Philippines) provides micro-insurance solutions to people without protection and access to financial assistance when faced with accidents, climate-driven calamities, diseases and other emergencies. It optimises AI and real-time data analytics for customer service and underwriting.

Map by Alliance4AI and Briter Bridges

Artificial Intelligence: The Promise and Peril in Developing Markets

The case for AI helping to address the SDGs cannot be overstated. However, as with other transformative and revolutionary technologies, there are structural challenges that can challenge the deployment of AI in development.

The IDIA Working Group on Artificial Intelligence & Development has highlighted some of the key challenges in deploying AI in international development:

  • Availability, accessibility and quality of data:
    A deeper, broader, and more accessible pool of data is needed to enable researchers, developers, and users to deploy AI solutions. Quality data is not always available or accessible in developing markets, particularly in fragile/conflict-affected contexts.
  • Capacity to engage with or use AI:
    A shortage of AI knowledge, skills and expertise is lacking globally. The expertise tends to be in the hands of a select few which is limiting its spread across more geographically and ethnically diverse groups. The dearth of AI-ready workforce is a huge challenge in developing markets.
  • Data ownership, privacy and security:
    A fair and inclusive use, and the security of personal data has to be critical issue. Unfortunately, there are currently insufficient regulations to protect against data misuse and personal privacy breach in most developing markets.
  • Governance, accountability and transparency:
    Efforts to define clear governance, accountability and transparency structures around the responsible use of AI have been very slow to materialise in developing markets.
  • Bias, discrimination and inequity:
    AI can help reduce human subjectivity and bias but it can also perpetuate and scale bias since the integrity of AI applications are dependent on training data that it learns from.

Stay tuned for more blog posts in this series and our upcoming publications where we aim to further explore the impact and scalability of AI applications in the developing world.