Barriers of AI adoption in low-income countries


With breakthroughs in cheap computing power, cloud services, big data, and advancements in machine learning (ML), artificial intelligence (AI) automates functions and enables new services worldwide. AI has the potential to transform and improve the way governments, organizations, and individuals deliver services, access information, and plan and operate.

Because it encompasses a wide range of business services solutions, from accounting and decision making to customer service, business intelligence and analytics emerged as a clear leader in using AI.

From sophisticated diagnosis and treatment options to hospital management systems, personalized lifestyle change recommendations, and healthy eating habits, healthcare is a leader and benefits from AI solutions.

Recent studies in Sub-Saharan Africa, North Africa, and South and Southeast Asia have found several barriers and challenges to implementing AI in low-income countries. India, Nigeria, South Africa, Egypt, Indonesia, Pakistan, Malaysia, Tunisia, Ghana, and Vietnam are involved.

This article will look at some of the major roadblocks to AI adoption in low-and middle-income countries (LMICs).

1. Availability, accessibility, and quality of data

To develop and train effective algorithms, AI and ML applications require a lot of data. There are comparatively fewer data available in less digitized environments, and less effective data practices can result in data held by companies and other organizations being uneven and inaccessible.

2. Access to reliable and affordable internet

Poor internet connectivity in both urban and rural areas prevents consistent use of mobile apps and consumer adoption of AI-based services. In Africa, for example, an estimated 267 million people and 53 million households do not have access to the internet. There are significant access disparities between urban and rural populations, men and women, youth and older adults, and higher and lower-income groups. While this has important implications for gender and inclusion, it also means that AI-based services and solutions in LMICs may lose market share. The high cost of mobile internet data or a home-based broadband connection also limits the size of the market and the number of people who use it.

3. Lack of access to sufficient computing power

In LMICs, a lack of sufficient computing power is a significant impediment to homegrown AI innovation. While cloud computing is a significant step toward overcoming this barrier, the cloud’s impact is limited in many areas due to unreliable internet.

4. Digital inclusion and connectivity: device access, ownership, and capability

AI-based solutions can benefit from data-rich, always-on apps that are updated regularly. However, these require adequate and consistent connectivity, which is difficult to achieve in areas with limited internet access, such as much of rural Africa and Asia. Poorer areas and groups are also left out of data-intensive services, where the cost of mobile data is a major stumbling block.

In LMICs, there is also a lack of smartphone penetration. Although smartphone penetration in Sub-Saharan Africa and South and Southeast Asia was 45 percent and 64 percent, respectively, in 2018, and is expected to rise to 67 percent and 81 percent by 2025, many of these devices are older models with operating systems that may not support high-tech apps. These limitations can be alleviated through thoughtful design, such as online/offline functionality.

5. Unreliable power infrastructure

Unreliable power generation and transmission plague many LMICs, resulting in frequent power outages and fluctuations. This can be a significant impediment to running powerful computers and a barrier for customers who rely on power to access the internet and AI-based services.

6. Human capital, education, and skills

While access to AI upskilling and training is expanding, many countries still lack a consistent homegrown and skilled AI developers pipeline. From preparing data and training machine learning to developing, launching, and maintaining new apps and services, deploying homegrown AI-based services requires specific skills. In many countries, the lack of mentorship for start-ups developing AI-based solutions is also a barrier. Digital start-up incubators and accelerators in LMICs vary greatly in quality, experience, and effectiveness. Users’ literacy and/or digital proficiency, which is required to access and use an AgriTech app, can also be a significant barrier to service adoption.

7. Lack of investment

Typically, AI-based solutions necessitate a significant investment. Unlike countries like China and the US, most LMICs have limited investment and funding. India, Kenya, Malaysia, Thailand, and South Africa appear to have higher levels of investment in Africa and South and Southeast Asia.

8. Poor transferability

AI applications created in other countries, businesses, and technological environments may not translate well and fail to produce similar results in different contexts. Scale, commercial viability, and language and climate differences (important for AgriTech solutions) will likely render AI-based solutions useless unless they are redesigned or recalibrated with locally sourced and reliable data. This is especially true when transferring solutions from technologically advanced countries to low- and middle-income countries (LMICs), where recipients may lack the awareness or voice to challenge low-quality AI solutions.

9. Automation and the risk of job losses

Uncertainty and fear can interfere with adoption, especially when understanding the role of AI and automation. Employees may be fearful of losing their jobs or resistant to change. Change management is critical within organizations to ensure that AI is implemented successfully. Other roadblocks include a lack of computing power, a lack of investments, and a lack of transferability.