Encountering the AI revolution: the role of development cooperation

Even though artificial intelligence (AI) has, knowingly or not, become interwoven into the fabric of our everyday lives – from tailoring advertisements to unlocking our phones – the latest developments in generative AI are a game-changer. As tasks that once required human skills are effortlessly supported or completed by AI models, reactions range from awe to worry.

AI for the Sustainable Development Goals: a considerable potential

There is ample reason for excitement from a development cooperation standpoint, as the potential to accelerate the Sustainable Development Goals (SDGs) is considerable. AI might support the work of teachers in remote areas where access to education is sparse. In regions with a high shortage of doctors and limited availability of healthcare services, AI technologies can potentially enable faster diagnoses or supporting treatments. GIZ has already been utilising AI within various projects with partner countries globally, from identifying and tracing illicit financial flows to detecting plant diseases and supporting international negotiations relevant to the SDGs through real-time analysis of negotiating positions, providing a stronger evidence base. In general, if used wisely, AI can fight poverty and improve equality worldwide and for women.

Do no harm with AI? The main challenge is to do better

However, the challenges associated with AI are considerable as well. Pessimistic predictions have described apocalyptic scenarios in headlines suggesting that AI will take over the world with the potential to harm humanity severely.

While this remains controversial, it is important to recognise the short- and medium-term risks of AI:

• Democratic processes: as evidenced by various examples, democratic processes could be threatened by easily generated disinformation, which may amplify political tensions or societal fragmentation.
• The future of work: the advent of generative AI is transforming the labour market, potentially leading to job losses and changes in job profiles. It could create new divides between those who can understand and utilise this technology and those who cannot. Given that there are disparities in access to education and to technological devices in our partner countries, with considerable disadvantages for women and girls, AI might increase existing inequalities. Even before the emergence of generative AI, concerns have been raised over poor working conditions for data labellers in developing countries, akin to ‘digital sweatshops’. And there are numerous instances where women, in particular, face disadvantages in AI-driven job processes, such as automated candidate pre-selection. This programming occurs as AI systems replicate past discriminatory patterns present in the data on which they are trained.
• Climate footprint: AI has also faced criticism for its significant environmental impacts in terms of energy consumption and resource depletion, often at the expense of low- and middle-income countries. Evidence suggests that women and girls are disproportionately affected by the climate crisis, which could exacerbate the existing their existing detriment.
• Digital infrastructure: access to models and computational power remains concentrated in high-income countries, widening (digital) inequality. This inequality may impede a country’s ability to maintain control over its data and utilise it for its benefit, endangering data sovereignty and technological autonomy.
• Non-representative training data: large language models, the foundation of generative AI solutions, must be trained with vast amounts of text, which is never neutral. The training data for these models are predominantly sourced from specific regions of the world, in specific languages, reflecting their values and norms. This setup exacerbates existing inequalities and biases, potentially excluding or underrepresenting certain groups due to historical marginalisation or structural discrimination and even spreading and reinforcing negative stereotypes. And with women being the biggest marginalised group worldwide, this is threatening the achievements in gender equality of the last decade.

The way forward: a people-centred approach

Development cooperation needs to address different layers:

First, to continue exploring and utilising AI tools to achieve the SDGs. Over the last few years, GIZ has increasingly used AI to fight poverty, reduce inequalities and achieve a just transition in climate change and agriculture in collaboration with partner countries. Over 20 AI-based pilots have been developed worldwide, and it is crucial to scale up these initiatives in collaboration with our European and international partners.

Second, to support regulation while finding the right balance. A regulatory framework on the national and international levels must prioritise safety and fundamental rights. There is reason for optimism about the European Union’s AI Act, the first-ever comprehensive legal framework on AI worldwide, which provides clear requirements and obligations for specific uses of AI. As such, it can be an example of ensuring trustworthy AI rooted in human rights while fostering innovation. So, how can we best support our partner countries in becoming active players in shaping their own governance of AI nationally? International development cooperation organisations must first listen and understand the individual needs and support them in unleashing AI’s economic and social potential while preventing harm associated with the mass collection and processing of personal data. This can include exploring if the European example serves as a suitable inspiration for trustworthy AI rooted in human rights. International partnerships can share best practices, promote adapting AI to local conditions and especially support market access to the EU. In that spirit, German development cooperation has already supported three political frameworks based upon European blueprints.

Next,responsibility in AI must be a guiding principle throughout. GIZ can build upon current initiatives regarding the vast field of responsible AI. For instance, on behalf of the German Federal Ministry for Economic Cooperation and Development (BMZ), the ‘FAIR Forward – AI for All’ initiative promotes policy frameworks for responsible AI and supports assessing the ethical impact of AI systems. The initiative enhances grassroots innovation by providing access to open-source training data, fostering the development of AI solutions rooted in the specific realities on the ground. Over 77,000 participants have engaged in e-learning courses on AI and more than 10 multi-country events have been organised to bring together various AI stakeholder groups. International cooperation should strive to amplify these efforts and comparable initiatives, bolster fair and open AI ecosystems on the ground, build capacity, support digital infrastructure and close existing data gaps.

Last, AI Feminism means more than just supporting women. With AI’s significant impact on women and girls, there is a pressing need for international cooperation to adopt a comprehensive feminist AI policy. This goes beyond merely supporting women and girls in the AI sector; it entails adhering to feminist principles throughout every stage of the data value chain. A feminist approach to AI involves analysing data’s inherent power structures, challenging binary thinking and hierarchical systems. It advocates for embracing pluralism in AI and data, with a focus on prioritising local and indigenous knowledge. Additionally, it raises questions about data ownership, collection processes and decision-making authority. Central to this approach is the promotion of participation and inclusion, giving women a voice in the predominantly male-dominated tech industry.

Complex AI ecosystem: the need for a strong development cooperation

The AI ecosystem is complex and every actor has a different role to play. The role of development cooperation is key – as a convener and honest broker for equal rights, as an ambassador for addressing the specific challenges for women and girls, as an advisor for regulatory frameworks and local innovations, as a capacity builder for systems and enablers and awareness raiser for the risks and potentials. Given the myriad layers, challenges and opportunities the latest AI revolution brings, this will be a marathon, not a sprint.

The views expressed in this #CriticalThinking article reflect those of the author(s) and not of Friends of Europe.