Digital transformation: From strategy to implementation

On a Wednesday morning in the middle of this industrial revolution, Jessica checks her smartphone. One click, another job. The start of the next twelve-hour workday in São Paulo, Brazil. Jessica, 24, shoulders a bag and boards a bus to downtown – as one of 600,000 workers in Brazil alone who take their daily orders via digital platforms. Jessica works as a cleaner. “The platform keeps half of the job amount as a service fee,” she says, but at least she has money in her pocket at the end of the day. “For jobs in richer areas they pay more; for me, the hourly wage always stays the same.”

Her work involves several gigs a day – short-term jobs as a freelancer. According to forecasts, by 2025, one-third of all job opportunities worldwide will be on digital platforms; the gig economy is growing rapidly. In low- and middle-income countries, these platforms lower the barriers to entering the labor market. Sometimes, as in Jessica’s case, the client is located in the same city, sometimes on another continent. For example, behind the chatbot ChatGPT there are occasionally thousands of clickworkers who feed and train the artificial intelligence for their voice messages at an hourly wage of two US dollars. The gig economy is creating new jobs, new incomes. On the other hand, digitalization is also bringing existing social inequalities into its own fields of work, widening the gap between rich and poor. Workers like Jessica are self-employed on the platforms: those who don’t accept a job make room for others who do. Massive asymmetries of power are prevalent, legal duties of care are lacking; the precarization of work is taking its course.

One could also do otherwise. Standards exist for trade or textiles, but not for labor. The Gig Economy Initiative of the German Federal Ministry for Economic Cooperation and Development (BMZ) is working on this. Among other things, it supports the “Fairwork Foundation” founded in 2018 by the Oxford Internet Institute, which aims to make working conditions on platforms more transparent and improve them. “These services are not going to disappear,” says Mark Graham, Oxford professor of Internet geography and director of “Fairwork”: “It’s unethical to say either we do this with an exploitative, underpaid and dangerous job or we don’t do it at all.” In cooperation with the BMZ, “Fairwork” tries to understand the workers’ conditions and evaluates the platforms according to developed principles; it has entered into a consultation process with some of them and achieved success: “Fairwork” is now active in 38 countries. “We’re not making utopian demands of them, but showing them how important minimum fair work standards are actually being implemented – often by their competitors,” Graham says. “Some of the platforms recognize the value we can add to their business and choose to work with us proactively.”

Fair labor is one of many components of the BMZ‘s digital strategy. Digitalization is not only integrated into development policy processes in order to achieve stated goals more effectively and transparently. It simply cannot be circumvented and is instead a tool of the present for shaping the future. The tracks are being built now, and those who want to help determine the direction must act. Digitization is being inscribed in the DNA of development cooperation, not just in the 600 current projects in 90 BMZ countries that are already using digital technologies, but in the structures themselves. Codewords of this BMZ digital policy are “value orientation” and “socio-ecological-feminist digital transformation.” Why? This is not only a question of justice, but also of efficiency.

The initial situation: 2.7 billion people are still offline, and only one in five women and one in three men worldwide have access to the Internet. Development work has recognized for many years that efforts without a focus on women and their rights are not very promising; they then miss 50 percent of the population. So why give up half of the skills? No progress without women – this fact can be denied, but it cannot disappear. For the BMZ and the GIZ as a service provider, this means that projects on these digitalization topics take longer, require more staying power, as more challenges are dealt with simultanously. Meanwhile, the so-called “return on investment” pays off immediately: If the global community does not want to lose sight of the sustainability goals by 2030, it needs digital jumps in development. And for that, it needs all people.

Open Source projects

New approaches and methods in international collaboration are also the focus of a creative space whose furnishings are a single message. Innovative office design, modular furniture, modern IT. Today, the “digilab” operates on the first floor of the ministry in Berlin Mitte – a kind of agile project group between BMZ, GIZ and KfW for digital-transformative consulting services. “The space is disruptive,” says Manuel Marx, “there is more of a start-up spirit here than that of a government agency.” The GIZ project manager is on site following the latest planning status of “Mojaloop,” an open-source digital payment infrastructure that is being jointly funded with the Bill & Melinda Gates Foundation. A glance at his screen shows rattling rows of numbers and then a virtual whiteboard where Marx is studying the requirements of various users. “Mojaloop is about inclusive networking of different financial partners, from smallholder farms to banks. It enables the creation of an inclusive service system for an entire country or region, helping local people access and use digital financial services according to their needs.” In many lower-income countries, people are unbanked. Digitalized payments can act as a driver for financial inclusion and social change. “Mojaloop is digital public good with enormous potential,” Marx says. Today, he will prepare another call with local implementation partner AfricaNenda, the company working on the technological setup of digital financial systems in many African countries and preparing to launch Mojaloop in Rwanda. “The setup has been created,” he says with satisfaction. “Now it’s on to implementation.”

Open source is a basic requirement not only for this project. The BMZ is pursuing a strategy of digitalization that is controlled neither by the state nor by the market, but instead pays attention to data sovereignty. The big players such as the USA and China set other priorities. They dominate the global field of digital value creation. Accordingly, they determine rules. This makes it all the more important to create spaces on the Internet and in artificial intelligence (AI) that are open to all.

Democratization of artificial intelligence by breaking down language barriers

One barrier of many: language. In African countries, for example, many applications are not in local languages; not everyone speaks, reads, or writes English or French. Some BMZ projects address this underrepresentation of African and Asian languages in AI development. Digital language services in particular lend themselves to inclusion. One example: a chatbot in Rwanda that provides detailed information about COVID-19. A cell phone, enter a short sequence of digits, and you can find out in the local language, Kinyarwanda, how many cases there are in the region, where you can get tested and what to look out for when you’re on the road.
GIZ’s FAIR Forward project developed the chatbot together with the Mozilla Foundation and local partners such as the start-up Umuganda.

And this is just the beginning: More than 2000 hours of speech data have been recorded and processed in Kinyarwanda so far. The data is freely and openly available to all developers. This has created a chatbot prototype, a text-based dialog system. “Kinyarwanda has become the fastest growing dataset and second largest open voice dataset in the world,” says Audace Niyonkuru, founder of Digital Umuganda. “Umuganda” is the name of the day each month when people in Rwanda do volunteer work. The startup has brought this day into the digital age, regularly calling on people to “donate” their language so others can benefit. In the process, the phrases in Kinyarwanda were recorded by thousands of volunteers. The language is spoken by more than 12 million people, including neighboring states. This training data promotes the entire local digital environment. They allow developers from other regions to access it and adapt the information for regional purposes. This is because the data collected has many uses, from interactive citizen engagement to apps that identify plant species and diseases, to chatbots that answer questions about sustainable agriculture, for example. Open data collection promotes digital innovation, especially in the Global South, and thus contributes to the democratization of artificial intelligence.

These new collections also bridge the classic childhood diseases of AI, namely the information gaps in the collection of data that adequately depicts the realities of life for all genders. Otherwise, the existing marginalization of certain groups continues online. As algorithms are trained on existing data, they reproduce existing biases and reinforce discrimination. AI systems are shaped by the priorities and biases – conscious and unconscious – of the people who develop them: a phenomenon Joy Buolamwini calls “coded bias.” “Women and BPoC are immensely underrepresented in AI development,” says the founder of the Algorithmic Justice League initiative. Algorithmic audits are considered a key strategy for uncovering algorithmic bias. “On the road to algorithmic fairness and transparency, we are still at the beginning. There are still many doors open to us in terms of shaping the future of artificial intelligence. But if we want to make it human-centered, inclusive, and feminist, we must act now.”

A simple example with ChatGPT shows how much room there is for improvement in current AI applications. “If you ask ChatGPT to describe a traditional wedding, you get a narrative of the bride in white and the groom in a black suit,” says Kathleen Ziemann, co-lead of the “Fair Forward – Artificial Intelligence for all” project at GIZ. “But this portrayed view has little to do with the reality of, for example, an Indian wedding.” There are many such examples of the cultural bias of data sets. The project is dedicated to the open and sustainable development and use of artificial intelligence, supporting partner countries in Africa and Asia in particular.

Sustainable transformation through artificial intelligence

Another generally overlooked aspect is the ecological footprint of AI. A Google search query, for example, requires a computing power of 0.3 watt hours, which corresponds to the consumption of an energy-saving lamp that burns for one hour. ChatGPT takes this into another dimension: A search request there consumes 1,000 times as much power as the Google search above and corresponds to 60 cell phone charges; it is also true for AI that power does not only come from the wall socket. So, just as the BMZ brings together two things with “fair” and “feminist” that have traditionally been thought of separately, German development policy now combines ” digitalization” and “climate”. Together, the two blocks can make a real difference: the digital and the green transformation toward a climate-friendly future – within the natural limits of our planet. Together, they form the Twin Transition. Many projects are still in their infancy, just getting started. Twin Transition is the biggest driver of change, especially in development cooperation.

The commitment of the Digital Transformation Center (DTC) in Indonesia, a GIZ institution, shows how AI is driving existing green transformation projects. The DTC is an institution of the GIZ, which among other things aims to protect the planet and preserve life on earth. The island nation’s enormous rainforests are a bastion against climate change, as they store large amounts of carbon. However, these forests are shrinking. Their area is in high request for agriculture, and palm oil in particular is in strong global demand. Clearances are the result, accompanied by legal disputes over who owns the land; not rarely, indigenous inhabitants are displaced.

In order to bring in knowledge at this point, areas are mapped using reference field data by involving the local mapping organization “Indonesian Community Mapping Network” or “Jaringan Kerja Pemetaan Indonesia” (JKPP) and the “High Carbon Stock Approach (HCSA) Foundation“. The project is financed by “FAIR Forward“, supported by DTC Indonesia.

The focus of this mapping is on the ability of forests to store carbon: The denser the green, the more intense the emission block. “Accurately mapping these areas and its biomass is an important step,” says Ruth Schmidt, a consultant with GIZ‘s “FAIR Forward” initiative. “This will allow informed decisions to be made about which areas should be protected from deforestation and to preserve indigenous habitat, and which are suitable for sustainable agriculture to promote sustainable supply chains with smallholder farmers.” The accuracy of maps is necessary for effective land use planning and management to support sustainable forestry practices, including responsible sourcing of products”.

And this is where AI comes into play: The cooperation partner ETH Zurich uses the collected data together with high-resolution satellite images as a basis to generate computer models with a resolution of up to ten meters. As such, AI makes this approach to open data collection more efficient, cost-effective, and offers the ability to scale across geographic and regional boundaries. “Forest in Indonesia has been recognized its role in storing carbons for the world,” says Yustina Artati, Senior Research Officer of the Center for International Forestry Research and World Agroforestry (CIFOR-ICRAF), based in Indonesia. “Government data shows deforestation has declined sharply in recent years.” Forests also play an important role in the livelihoods of many people, the researcher said. And her colleague Beni Okarda: “AI can function like a warning system for the state to know where to respond quickly to protect forests.” AI will also play an increasingly important role in other areas of disaster management, such as forest fires or tsunamis, says the Senior Research Officer at CIFOR-ICRAF.”

AI usually shows two faces. On the one hand, it helps farmers use simple apps to detect diseases in their crops in real time or to classify agricultural crops using local data sets; it enables illiterate people to participate via voice control, boosts the development of innovative products and it can break down barriers in developing and emerging countries. But on the other hand, there is a lack of frameworks for the responsible and locally adapted use of AI: today, for example, only 25 of 54 African countries have data protection legislation and only a few pioneering countries such as Kenya and India have an AI strategy. The “FAIR Forward” initiative is committed to empowerment around value-based AI, including building local learning capacity, providing local training data, and open-source AI technologies.

Digitalization is a human project. Whether we like it or not, we are on an unstoppable journey into the future. And we decide which waypoints emerge on this journey. The digital strategy of German development policy offers guidance toward sustainable growth, mitigating threats to humanity and the planet – and a better world for all.