1 Introduction

The digital divide refers to the disparity between countries, regions, and people in their access to digital services (Fuchs & Horak, 2008; Van Dijk, 2020; Potter et al., 2008; Mubarak, 2015; Bon, 2020). For example, in 2023, 57% of the African population is unconnected to the global digital society (Internet World Stats, 2023). This disparity is considered a global challenge. It is obvious that digital information and communication are of key importance for the development of countries and regions.

Different approaches exist to address the digital divide. Some approaches are technology-focused, seeking progress through technological advancement. Some approaches focus mainly on profits and economic growth. Other approaches are more human-centric, trying to achieve human well-being. Each approach reflects a different worldview or school of thought, with varying assumptions about the nature of development, the role of technology, and the desired outcomes of bridging the digital divide.

From a technology-focused perspective, not being connected is attributed to a lack of physical and digital infrastructure. Due to the absence of proper infrastructures in poor regions or countries, many people do not have access to digital services and content. This withholds their participation in the information society and hampers development. Rolling out digital infrastructure and making the Internet available to people in every corner of the planet is then the solution. The ensuing question is how to organize this.

From a slightly different perspective, the digital divide is only a snapshot in time, in an ongoing process of technological innovation and diffusion (Rogers, 2003). Poor regions are lagging but will soon catch up in their technological and organizational capacities and will adopt digital technologies at a later stage. This will lead to progress and bring economic development, at long last (De Janvry & Sadoulet, 2021; Taylor & Lybbert, 2020). The uptake of digital technologies by people in low-resource environments is seen as an opportunity, as this will create a new market segment for technology vendors and other markets. This market segment is referred to as the “bottom of the pyramid” (Prahalad, 2004; Heeks, 2008). Economic growth is the main justification for this approach.

Both perspectives—technology-focused or economy-focused—have one idea in common: digital technology is the motor for progress and development. Access to digital information and services will lead, from an “underdeveloped” situation to a situation of prosperity, like in the industrialized countries (Wicander, 2009). It is assumed that well-being will follow from technological and economic development and that prosperity will trickle down, naturally, to most people.

Alternative perspectives on the digital divide value a more human-centered standpoint. Digital humanism, decolonial theory, and grassroots initiatives are examples of human-centered approaches. They prioritize human well-being, social inclusion, equity, and emancipation above technology and profit.

In 2019, the Digital Humanism Initiative was launched in Vienna by a group of primarily European scientists, policy makers, and practitioners concerned about the disrupting impact of digital progress on society, economy, and the natural environment. The Vienna Manifesto for Digital Humanism was proclaimed as a call for ethics and humanism and as an antidote to technocratic approaches with their focus on merely technological innovation and economic growth (DIGHUM, 2019). The digital humanism movement is now supported by many scholars, politicians, and policy makers from around the world. It launched a debate about the purpose and the future of the digital society (DIGHUM, 2019). At the onset, the Digital Humanism Initiative focused on the societal issues of the digital society in the Global North, where it originated, but soon it embraced a cosmopolitan viewpoint, in which the Global South and the exclusion of people in low-resource environments are a central problem to be discussed and addressed.

A similar, and even more radical, perspective is found in decolonial studies, in which patterns of inequality, poverty and wealth, domination, and colonialism are explored and analyzed in society (Mignolo & Walsh, 2018; Mendoza, 2021). Decolonial theory offers an alternative lens to assess the digital divide and explore its root causes and origins, in the light of colonialism. The conjecture here is that the digital divide is inherently unfair because it excludes people. However, the idea also exists that digital technologies, if well designed, may be able to help improve certain aspects and overcome certain problems of people. Of course, these digital technologies must be, by design, respectful of human rights and human dignity and of local agency, culture, norms, and values.

Whereas academic studies, e.g., Lin et al. (2015) and De et al. (2018), express criticism about various aspects of the digital divide, only a few of these studies have been action oriented. Yet, from a digital humanism perspective, the purpose and goal are not only to discuss but also to influence policy, raise awareness, and call for practical, problem-solving action. An example could be an interdisciplinary study that combines action research, design science, and a human-centered approach to design technologies that support people and communities in low-resource environments. The persistent question is how this study can ensure fairness, ethics, and inclusion by design while at the same time being respectful at the autonomy and agency of local users, communities, and people.

This question will be discussed in the sections that follow. In Sect. 2, we give a brief history of the digital divide. In Sect. 3, we observe the digital divide through a decolonial lens, assessing roots, causes, and patterns of inequality. In Sect. 4, we seek the requirements for inclusive, human-centered design. In Sect. 5, we propose a human-centered methodology that combines action research, design science, and ethics perspective. We illustrate this through a case study from northern Ghana in Sect. 6.

2 Short History of the Efforts to Bridge the Digital Divide

Since the 1990s, many efforts have been undertaken by the international communityFootnote 1 including international organizations such as the World Bank, the United Nations, UNESCO, national and international development agencies, numerous non-governmental organizations (NGOs), and large private charity organizations to make digital technologies available for social and economic development of poor regions of the world. Despite large budgets and numerous projects in prestigious international development programs, a mismatch between technologies and the target environment often occurred (Bon, 2020). Many unsuccessful technology transfers resulted from blind optimism about the impact and reach of digital technologies, combined with poor understanding of local needs and context. Techno-optimism about digital technology culminated in the United Nations World Summit on the Information Society, in 2005, in Tunis, where heads of states from all around the world and global players in the big private technology sector came together, including Microsoft, HP, Nokia, and Intel (Berry, 2006; Qureshi, 2006). Goals were set during this summit, to roll out the Internet before the year 2020, in every corner of the planet.

Soon after 2005, it became clear that many ambitious, large-scale development projects were failing, especially in serving the poorest and least connected communities with digital technologies (Bon, 2020). Two examples are (i) the One Laptop Per Child (OLPC) project, which started around 2008 and aimed to provide every schoolchild in the world with a laptop (Buchele & Owusu-Aning, 2007), and (ii) the Millennium Villages project, which started in 2005 and selected a small number of very poor and unconnected villages across African countries to pull them “out of poverty” by giving them an investment boost, transferring state-of-the-art Western technology, and connecting the communities to high-speed Internet (Sanchez et al., 2007). Both the OLPC and the Millennium Villages project spent large sums but did not meet their objectives or expectationsFootnote 2. The failures of these and many other large-scale technology transfer projects tempered the initial techno-optimism of the international community.

Unexpectedly, it was not the Internet, but simple, mobile, voice-based telephony, rolled out by the telecom companies in many remote and poor regions to low-income communities of the world, that became a huge success and had a large impact for people’s lives at the so-called base of the pyramid (Kalba, 2008). For many poor people in urban or remote rural areas in developing countries, mobile telephony was their first digital experience and remote communication. The mobile telephony success was followed by an innovation called mobile money. This technological innovation, which only required a simple mobile phone—no smartphone or Internet connection—enabled people to deposit money on their cell phones and transfer it to other users of the service, without need for a bank account. Mobile money was first launched in 2007 in Kenya, as a truly African innovation by the local telecom operator Safaricom. This service registered 1.1 million users in the first 8 months after its inception (Mbiti & Weil, 2011). Soon, many countries in Africa followed, as Orange, Airtel, and many other telecom companies started providing mobile payment services to millions of previously unbanked citizens in Africa (Nan, 2019).

So, in the past two decades, the least developed countries in, e.g., sub-Saharan Africa have caught up very rapidly and entered the digital era. However, innovation and technology adoption often followed a different path than in the Global North. For example, many people in rural Africa skipped text-based systems and are now using asynchronous voice-based communication through voice messaging systems and social media apps, without ever having sent a written (text-based) email (Dittoh et al., 2020a, 2020b).

Despite the positive impact of mobile telephony and mobile money for many people in rural Africa, the profits remained unequally shared. At the onset of the 2020s, many sub-Saharan countries were mainly consuming digital services but hardly developing or producing any.

From the point of view of infrastructure, only few data centers or Internet exchange hubs exist in sub-Saharan countries (Augustine, 2022). The data produced by African users are stored in the cloud, i.e., remote data centersFootnote 3. The costs for data transport between users and the “cloud” make digital services more expensive in remote areas than, e.g., in urban centers. This is another example of inequality.

Technological innovation is moving fast. The technology leap in the second decade of the twenty-first century is based on data-driven artificial intelligent systems. Many new beneficial solutions are to be expected, e.g., for the development of speech recognition of local indigenous languages or for the digitalization and access to specific, contextual knowledge systems. Yet, big concerns exist, related to the deployment of generative AI.

People in countries in the Global South may be vulnerable to the spread of disinformation and digital surveillance. They may face infringement of their intellectual property rights, due to lack of legislation and governance of the digital sphere (DIGHUM, 2019). Since privacy rules and regulations such as GDPR are still not implemented in many African countries, there is no good protection of people’s identities, privacy, and data ownership. Especially people “at the base of the pyramid” may be less well informed and therefore more vulnerable to predatory technology firms and digital scams. Moreover, in the light of a rapidly involving technology like artificial intelligence, combined with a lack of policy and legislation in many countries in the Global South, this may result in a complete governance of the digital society, including the infrastructure, data, and technological know-how, to be concentrated in just a few current market-dominant private technological firms, also known as Tech Giants, such as Meta (Facebook), Alphabet (Google), Amazon, Apple, and Microsoft.

3 The Digital Divide Through a Decolonial Lens

The digital world, where we socialize, interact, and take decisions, mirrors the physical world in many aspects. This includes patterns of inequality and domination, which are often referred to as “coloniality” (Mignolo & Walsh, 2018; Mendoza, 2021 pp. 46–54; pp. 1–12; Quijano, 2007 pp. 15–18; Hope, 2022). Decolonial theory helps uncover hidden patterns of domination within social structures. In this discussion, we explore a few examples from the digital society.

One common assumption, also found in the SDG9, is that many underdevelopment-related problems, e.g., in education and trade, will be solved, once Internet connections have been established. Yet is this a valid assumption? In the light of the concerns about artificial intelligence and the increasing concentration of knowledge and power, it is debatable whether the Internet will bring democracy, social justice, equality, and a sustainable and prosperous life to all people (Bon et al., 2022).

Let’s take the example of “free Internet”, offered by big tech firms, such as Google (Dahir, 2020), Starlink (Sapah, 2023), and Facebook (The Guardian, 2016), to people in low-resource environments. These initiatives will ultimately connect many people, even in remote low-resource environments, to the digital backbones for free. However, we must question what “free Internet” means in a capitalist world. The revenues of the large technology companies rely on extracting value from personal data, people’s Web browsing preferences, by tracking their entire online behavior using AI algorithms. Users are often unaware of the exploitation of their personal data. This business model of extracting value from the people’s online behavior has been criticized by Shoshana Zuboff in her book The Age of Surveillance Capitalism (2019), for which she introduced the term “behavior surplus.”

Another aspect of decolonial critique is the hegemony of the Internet as a network standard. While the Internet can be seen as a global common, a platform to share information, access opportunities, and collaborate across geographic and cultural boundaries, as discussed earlier, it has disadvantages. Choosing not to be part of this network results in isolation, for the non-user. This characteristic of standards and networks, which is described by Grewal (2008, pp. 20–28), makes the Internet a hegemonic system without escape, despite the price users, communities, and even countries pay with money or data, to access it.

Geolocation and governance of digital infrastructures are also aspects related to coloniality. The digital society is heavily concentrated, physically, economically, and socially, in the Global North. The commercialization of digital technologies, influenced by this centralization, further channels the profits from innovations in the Global South to investors (Zuboff, 2019, pp. 63–96). Governance and decision-making in this realm are predominantly controlled by private tech firms, operating under norms and regulations from countries in the Global North. Unequal competition in terms of storage, connectivity, funding, and adoption also hinders start-ups in the Global South (Bon et al., 2022).

In many countries, big tech firms and telecom providers are assuming public roles and functions traditionally held by the state, such as in healthcare, education, and infrastructure. Through philanthropic gifts and corporate social responsibility, digital services are provided in exchange for data, market penetration, tax savings, branding, and policy influence. Youngsters are often targeted by big tech companies with media, music, video, entertainment, and news (Pini, 2020, pp. 37–40).

Coloniality is also evident in artificial intelligence (AI). Machine learning algorithms, previously assumed to be objective and value-free, have been found to harbor many discriminating biases (Mohamed et al., 2020, pp. 659–663). These biases are often concealed in the underlying data. Trivial examples include visualization programs that autonomously lighten the complexion of black and Asian faces or smart doors that, based on facial recognition algorithms, only open for white faces while remaining closed for a person with dark skin, as the machine learning algorithm fails to recognize them as a person. These racial biases can emerge unexpectedly in autonomous smart systems. If these algorithmic biases are not properly addressed, they will further perpetuate inequalities and injustice against certain groups and individuals (Mohamed et al., 2020, pp. 659–663).

The aforementioned discussion highlights the concerns regarding coloniality, domination, biases, and injustice. If left unaddressed, these issues will result in a digital society that is exclusive, unethical, and, from various perspectives, reminiscent of neo-colonialism.

It is crucial to recognize that innovation in digital technologies is not an autonomous process, but one driven by deliberate choices. Methodologies from information systems engineering provide flexible and powerful approaches for designing and fostering open innovation. Action research with its long-standing tradition of improving real-world situations adds to it the human-centered and ethical aspects. In the following section, we will delve into the discussion of how information and communication technologies can be designed in a human-centered manner.

4 Requirements for a Human-Centered Approach

When exploring alternative approaches to serve communities and individuals in low-resource environments, which principles are required?

The foremost and crucial principle is that digital technologies should not cause harm to anyone. It is imperative that digital services align with the objectives and goals defined by the local people and users. To achieve this, local users must actively participate in decision-making processes regarding the goals and objectives of the digital service.

Secondly, digital technology must be adaptable in a flexible way to the local context, enabling it to overcome local barriers. These barriers can include language, literacy, limited infrastructure availability, local purchasing power, or other context-related issues.

Thirdly, there must be local ownership of data and the protection of local domain knowledge and intellectual property. This is essential to prevent compromise or misuse of these valuable assets. In the case of co-design, ownership of the designed artifacts should also be shared. Research data should not be automatically owned by the researcher or published without permission or consent.

In summary, the development of digital services for unconnected people or communities in low-resource environments should be conducted in collaboration with local users. Since software developers and engineers often lack familiarity with the low-resource locations in question, significant emphasis is placed on communication between users and developers through collaborative workshops and co-design sessions.

In the next section, we will describe a methodology and approach that was developed in the field from 2009 to 2019 through an action research program named W4RA—the Web Alliance for Regreening in Africa (W4RA, 2023). The program’s objective is to support local smallholder farmers in three West African countries, Mali, Burkina Faso, and Ghana, by designing digital technologies to help them achieve their objectives. The approach is referred to as Information and Communication Technologies for Development, version 3.0, abbreviated as ICT4D 3.0 (Bon, 2016).

5 Combining Action Research, Design Science, and Ethics Perspective

ICT4D 3.0 represents a grassroots approach to bridging the digital divide, offering an alternative to the prevailing, economic growth-oriented approach to bridging the digital divide. This incumbent approach is described by Richard Heeks (2008) as “a new opportunity for ICT vendors to harness digital innovation and serve the world’s poor, profitably,” in an article, titled “ICT4D 2.0: The Next Phase of Applying ICT for International Development.

ICT4D 2.0 was presented by Heeks as a great improvement to the previous approaches, in which off-the-shelf digital technologies were transferred to poor regions, often with limited success (2008, p. 26).

In contrast to its predecessors, ICT4D 3.0 positions itself as a human-centered, grassroots methodology for designing digital technologies based on users’ needs. It follows a five-step framework:

  1. (i)

    Context analysis—understanding the users’ environment

  2. (ii)

    Needs assessment—understanding the user’s goal

  3. (iii)

    Use case and requirements analysis—defining and specifying a feasible digital solution

  4. (iv)

    Prototyping, engineering, and deployment of a digital service

  5. (v)

    Sustainability analysis—exploring the local business ecosystem to ensure the long-term availability of the service

ICT4D 3.0 is a goal-oriented approach, striving for a certain improvement for its envisaged users. A conceptual visualization of this approach can be represented as a goal-strategy map (Fig. 1, cf. (Rolland, 2007)), where each colored ellipse signifies an intention to be achieved through corresponding actions. The process is iterative and adaptable, allowing for adjustments based on new information.

Fig. 1
A diagram illustrates the five-step framework. They include assessment, use case and requirement analysis, prototyping, sustainability analysis, and context analysis.

ICT4D 3.0 as a collaborative, iterative, adaptive approach to digital development in low-resource environments. From: Bon, A. (2020, p. 204, Intervention or CollaborationRedesigning Information and Communication Technologies for Development, Amsterdam: Pangea)

Emancipation, autonomy, and inclusion are the core values ICT4D 3.0 strives to achieve. It has undergone continuous evaluation and validation by its users in the field (Bon, 2020). Importantly, it can be considered a decolonial approach as it empowers users to make decisions rather than having external parties impose solutions upon them.

6 Tiballi: A Case Study of AI and Data Science for Farmers in Ghana

A case study that illustrates the ICT4D 3.0 approach in a community-oriented, transdisciplinary research project is Tiballi, a research project, set up in 2023 in northern Ghana (Tiballi, 2023; Dittoh et al., 2021). Tiballi explores how artificial intelligence—and more specifically machine learning and natural language processing—can serve people in a low-resource environment. Tiballi’s envisaged users are proficient in a local, under-resourced African language only; they have low incomes, low literacy skills, and limited access to the Internet.

Tiballi’s project goal aligns with an explicit wish of farmers in northern Ghana to have access to rainfall data, as this information is essential in rain-fed agriculture, but often not accessible. These wishes were collected during workshops and focus group discussions; see Fig. 2. In these meetings, it becomes clear that (a lack of) Internet access is not the only barrier for the community to use digital services. It is also important to provision locally relevant information in the language and modality people are familiar with, such as radio or simple voice-based phones.

Fig. 2
A photograph of a discussion among a group of people in Nyankpala, Ghana.

Tiballi project’s needs assessment, during a focus group discussion in the village of Tingoli, community of Nyankpala, Ghana. Photograph by Gideon Amakama Ali (18 February 2023)

In the Tiballi project, rainfall data are collected in the field and combined with global weather and climate information available on the Web. The combined result is made available to the local community members in their language, through voice-based phone access or local radio (Baart et al., 2018). Automatic voice response and interaction is constructed from vocabularies which are collected in the community and processed with AI. Figure 2 shows a focus group discussion with farmers in the Tingoli village in northern Ghana.

In short, to deliver the requested information to the community, the Tiballi project aims to:

  1. (i)

    Collect and store real-time weather, for example, the cumulative rainfall over the season.

  2. (ii)

    Develop an automatic speech recognition system in the local language Dagbanli. It uses a previously developed method to resource small languages, in which a machine learning model is trained with a relatively small, locally collected dataset.

  3. (iii)

    Develop a voice-based application that combines limited vocabulary text-to-speech and automatic speech recognition in Dagbanli language. This will enable users to interact with the system by phone and receive requested information, every day (Stan et al., 2022).

  4. (iv)

    Develop and deploy a voice-based platform, based on inexpensive, small hardware, to provide the service, also in the absence of Internet access and avoid high costs of broadband connectivity (Baart et al., 2019).

7 Discussion on Critical and Societal Issues

In the previous sections, we proposed a collaborative, iterative, adaptive approach to digital development, aligned with inclusion and an ethics perspective. Yet, this approach may have its limitations, for example, related to sustainability or costs of digital services. It is important to continuously evaluate and critically reflect before, during, and after the project period. Deliberation and discussions with its users and beneficiaries on their personal experiences and opinions are essential.

Technological innovation can be a rapid process. Huge developments are taking place in (distributed, cloud, edge) computing, social media, blockchain technology, Internet of Things, artificial intelligence, and data science. It is important to continuously reassess potential positives and negatives of technologies introduced in low-resource environments. We as researchers, ICT professionals, and global citizens must be aware of our responsibility, to care for the present and the future of the digital society.

From the ideas in this chapter, we propose to support the lives and work of people in low-resource environments through co-design. Yet, this small-scale approach needs dissemination to scale up, for example, by training young professionals at European and African universities and by bringing together a community of developers and contributors.

For a societally oriented community of developers, it may be challenging to deliver, in a sustainable way, competitive digital services to low-resource environments. The strong monopolization by big technology firms with technologically advanced services, and backed by big investors, puts up high barriers for the smaller parties. Yet, niches exist—especially in less wealthy regions—that are still unserved.

This endorses the idea that deliberate choices can be made toward a better and more inclusive digital society, guided by social engagement, and encourage human-centered innovation, quoting the Vienna Manifesto 2019: “calling for a Digital Humanism that describes, analyzes, and, most importantly, influences the complex interplay of technology and humankind, for a better society and life, fully respecting universal human rights”Footnote 4 (DIGHUM, 2015).

8 Conclusions

Whereas, from the previous discussions, it is evident that, from a macro-economic and geopolitical perspective, there are no easy fixes to bridge the digital divide as a problem of global scale between regions and countries, we propose to develop small-scale grassroots approaches. We stress the importance of collaboration, at all levels.

Firstly, at the international policy level—in platforms and think tanks such as the Digital Humanism Initiative—representatives of communities in the least connected countries and regions in the Global South must be included, to make their perspectives visible, create awareness of the existing problems, and inform the public debate about the needs of people in low-resource environments.

Secondly, at the level of research and education, collaboration and exchange of knowledge is needed between universities (north-south and south-south). Research programs must be developed that are context oriented. Students must be trained through community service-oriented research and technology development.

Finally, it is important to innovate in technologies in and for people in low-resource environments and develop transdisciplinary research methodologies. In the paragraphs above, we discussed community-oriented, collaborative action research and socio-technical software development. While the exchange of ideas between developers and users can offshoot innovation in low-resource environments in unexpected ways, it can also be a source of inspiration for new forms of collaborative knowledge production.

Discussion Questions for Students and Their Teachers

  1. 1.

    What does it mean when it is said that the digital space is colonized?

  2. 2.

    What are the problems when digital technologies are transferred from industrialized countries to developing countries, and what would be the alternatives to digital development?

  3. 3.

    What are the decolonial aspects of digital technologies that reflect patterns from the social and physical world? Think of examples that are not mentioned in this chapter.

  4. 4.

    What is the fundamental difference between the critical realism perspective and the action research/design science perspective on the challenges related to the digital divide?

  5. 5.

    What are the five main contextual challenges for the design of digital services, of the low-resource environment in the Tiballi case study in northern Ghana, and how does the given framework ICT4D 3.0 address these challenges?

Learning Resources for Students

  1. 1.

    Potter, R. B., Binns, T., and Elliott, J. A. (2008) Geographies of development: An introduction to development studies. Essex: Pearson Education. Chapter 1.

    This book gives, from a geopolitical perspective, an introduction to the digital divide between countries and regions in the world and describes the historical backgrounds in the framework of recent and contemporary history.

  2. 2.

    Zuboff, S. (2019) The Age of Surveillance Capitalism: The Fight for a Human Future at the New Frontier of Power. London: Profile books.

    This book provides a critical analysis of the business models used by big technology firms and how these affect people and the economy, creating unfairness in the digital society and the physical world.

  3. 3.

    Mohamed, S., Png, M. T. and Isaac, W. (2020) Decolonial AI: Decolonial Theory as Sociotechnical Foresight in Artificial Intelligence. Philosophy & Technology, 33, pp. 659–684.

    This paper analyzes the risks and biases of artificial intelligence algorithms through the lens of decolonial theory.

  4. 4.

    Stan, G. et al. (2022) ‘A Lightweight Downscaled Approach to Automatic Speech Recognition for Small Indigenous Languages’, in 14th ACM Web Science Conference 2022 (pp. 451–458). Doi: https://doi.org/10.1145/3501247.3539017.

    This paper describes a technical method to build AI-based automatic speech recognition for indigenous languages, using small data and energy-efficient methods, making it more affordable and accessible for people and communities in low-resource environments,

  5. 5.

    Dittoh, F., et al. (2021). ‘Tibaŋsim: Information Access for Low-Resource Environments’, in: Conference proceedings by Springer AISC. ISBN Number—2194-5357 Series. Doi: https://doi.org/10.1007/97898116.2377662.

    This paper describes a real-world deployment of an ICT system in rural Ghana, as an example of collaborative technology design and its challenges in low-resource environments in the Global South.

  6. 6.

    Baart, A., et al. (2018) ‘Ney Yibeogo—Hello World: A voice service development platform to bridge the web’s digital divide’, in WEBIST 2018—Proceedings of the 14th International Conference on Web Information Systems and Technologies, pp. 23–34.

    This paper stresses the importance of resourcing local languages to bridge the digital divide and methods how to do this.