Conference 2025

Isabel Günther

• Head of NADEL, ETH Zurich
• President GRP-Alliance

In her presentation, Grace Kagho explored how AI and digital twins can radically transform urban planning—especially in emerging and low-income regions where planning challenges are most acute. Drawing from personal experience and global urbanization trends, Kagho highlighted the high cost of poor planning decisions, including chronic traffic congestion, health risks, and long-term socioeconomic impacts. These issues, common in rapidly growing cities like Lagos, Bogotá, or Jakarta, call for smarter, more inclusive planning approaches.

Grace Kagho introduced digital twins as powerful, virtual representations of urban systems that simulate real-world behaviors, such as transportation dynamics and human decision-making. Enhanced by generative AI, these tools can process complex urban data and respond to natural-language queries from users—allowing both experts and non-specialists to interactively explore scenarios and forecast the impacts of planning decisions. This capability shifts planning from a slow, expert-driven process to a more participatory, real-time system accessible to communities and decision-makers alike.

The presentation emphasized AI’s role in bridging data gaps, enabling predictive simulations, and supporting collaborative planning in under-resourced settings. However, Kagho also stressed the ethical imperatives: addressing data bias, safeguarding privacy, ensuring algorithmic transparency, and including diverse perspectives in tool development. Concluding with a call to action, Kagho urged the scientific community to adopt proactive, AI-supported planning methods and invited collaboration on real-world applications, such as ongoing pilots in Ghana and Latin America.

AI generated Sketch-Note of the presentation.

In his keynote Kibrom T. Sibhatu emphasized the transformative potential of Digital Advisory Services (DAS) and artificial intelligence to support smallholder farmers in Sub-Saharan Africa. DAS are digital platforms and tools that provide farmers with tailored, timely, and location-specific information to support agricultural decision-making based on weather forecasts, pest and disease alerts, financial advice, crop insurance, input recommendations, and sustainable land management guidance.

Sibhatu pointed out that traditional agricultural extension systems often fail to meet the needs of farmers due to limited resources, top-down communication, and a lack of contextual relevance. In contrast, DAS can reach large numbers of farmers, especially young people who are increasingly connected via mobile phones and the internet. When designed inclusively and with local realities in mind, DAS can significantly improve yields, income, and resilience.

Drawing on research from the AgriPath project and the Farmbetter app, he showed that DAS can lead to yield and income increases of up to 75 percent. These tools help farmers make informed decisions, adapt to climate change, and manage risks more effectively. However, he also highlighted persistent challenges, including poor connectivity, low digital literacy, and limited trust in digital platforms.

Sibhatu described artificial intelligence as a powerful tool to enhance DAS by enabling personalized advice, improving service quality through continuous learning, and reducing costs. At the same time, he stressed the need for further research to understand who adopts these tools, how they are best delivered, and how to ensure they are inclusive and trustworthy.

He concluded that DAS and AI should not replace traditional systems but rather complement them. For these innovations to be sustainable and impactful, they must be co-designed with farmers, adapted to local contexts, and supported by strong collaboration across sectors.

AI generated Sketch-Note of the presentation.

In her presentation, Kebene Wodajo addressed the multifaceted challenges and ethical considerations arising from the integration of artificial intelligence (AI) into scientific research and international collaborations. The talk began by grounding the discussion in the broader context of research partnerships, noting that successful collaborations demand more than technical expertise—they require ongoing negotiation of values, cultural sensitivity, legal alignment, and mutual respect across institutional and geographic boundaries.

As AI technologies increasingly mediate research practices and even become outputs of research themselves, Wodajo argued that the ethical landscape of research becomes significantly more complex. In particular, large language models and other advanced AI systems raise questions about epistemic justice—such as who controls the production of knowledge, how diverse ways of knowing are represented or flattened, and how existing power asymmetries may be reinforced or replicated by AI-driven processes.

To address these concerns, Wodajo introduced the "AI Value Chain" framework as a practical and critical tool for evaluating the ethical quality of AI use in research. This approach reframes AI not as a neutral technological solution, but as a socio-technical system shaped by the interaction of human labor, institutional choices, material infrastructures, and governance mechanisms. The value chain includes software, hardware, labor, financial, and regulatory components, all of which contribute to both the positive and potentially harmful impacts of AI systems.

Kebene Wodajo emphasized that many ethical issues—such as bias, discrimination, and labor exploitation—cannot be resolved through technical fixes alone. For example, bias is not merely a data problem; it is also cognitive, institutional, and systemic. Illustrative examples included underpaid data laborers supporting large language models and the need for transparent, fair sourcing of data and models.

The proposed AI Value Chain approach involves five phases: identifying the AI system in use, co-mapping its full value chain, assessing its implications collaboratively, recognizing ethical risks and opportunities, and planning for long-term, sustainable responsibility beyond the lifecycle of individual projects or tools. Wodajo concluded by advocating for a critical yet constructive engagement with AI in research—one that safeguards ethical integrity and foregrounds inclusivity, justice, and accountability across all stages of scientific collaboration.

AI generated Sketch-Note of the presentation.

In this workshop, led by Dr Kristina Pelikan and Dr Simona Dobra from Swiss TPH, we explored the intersection between artificial intelligence (AI) and multi-epistemic partnerships. We explored how AI could facilitate inclusive research practices within various knowledge systems. The session highlighted the challenges and opportunities of incorporating AI into transdisciplinary research projects, particularly in contexts where indigenous, local, and scientific knowledge systems converge.

Through concrete examples, such as a project on the interrelation of animal-human health (zoonoses) from Swiss TPH in collaboration with Maya healers, participants discussed how different perspectives sometimes clash when addressing health-related issues. This demonstrated the difficulty that AI tools have in understanding and translating the diverse cultural concepts inherent in languages.

Key insights included the distinction between generative and discriminative AI approaches, focusing on language understanding, ethical considerations, and data privacy issues. Examples of AI initiatives such as AIMS and African AI Startups were presented to emphasise the importance of incorporating local languages into AI systems to ensure that these tools reflect the needs and perspectives of local communities.

A central discussion point was how to create AI tools for and from the Global South, while considering the ethical implications of language use, and the risk of technology perpetuating a new form of colonialism. The session also explored the importance of context in AI-generated predictions, highlighting the necessity of human judgement when interpreting AI results, particularly in contexts with diverse worldviews. In a nutshell, the workshop addressed how AI tools could be designed to better reflect the cultural contexts underlying languages.

Big data and data processing with artificial intelligence are increasingly seen as tools in support of sustainability transitions, for example in sustainability initiatives as the European Deforestation Regulation EUDR which states that commodoties imported into the EU should be deforestation-free. However, unprotected flows of big data bear risks of being (mis)used as a grid for surveillance, commercial exploitation, or even political influence. The European Union, China, and the USA have created three major data governance realms which their trade partners need to abide by. This invokes concerns over power imbalances and colonial legacies echoing into the digital era. In this workshop, we discussed questions around data sovereignty and its implications, and around data as a new form of capital.

A concrete example from the food value chain was given in the form of an app developed for smallholder farmers in Nigeria to monitor the quality of their produce stored in cold rooms in real time. The app combines a digital twin of fruit, and the actual measurement data of the cold room. Farmers can also see the current market prices on the app. The potentials and pitfalls of digital twins for sustainability transitions were then discussed with the workshop participants. Potentials of digital twins includes the ability to simplify complex systems and be able to experiment and try out different scenarios. Pitfalls include high energy use of systems using big data, and a problem of data bias and potential misinformation.

AI is starting to transform the way researchers gather, analyze, and interpret data. In this workshop, moderators Dr. Edward Asiedu and Dr. Yael Borofsky will kick off the conversation about the potential future effects of AI on local participation in research within developing countries. Building on these initial insights, the group will engage in an energetic discussion, considering viewpoints from participants and expanding the conversation to how AI will influence global research partnerships. The workshop's aim is not to reach a definitive conclusion but to encourage dialogue about the opportunities and challenges of AI-driven social research, ensuring that local perspectives remain integral to pro-poor research amidst technological progress.

We were pleased to use the recent conference as a platform to officially launch our new name: Swiss Alliance for Global Research Partnerships (GRP Alliance). This marks a significant moment in our journey.

Since our foundation in 1994, we have operated under the name Commission for Research Partnerships with Developing Countries (KFPE). Over the past three decades, the landscape of international research collaboration has evolved considerably. The binary distinction between "developed" and "developing" countries no longer adequately reflects the complexity and diversity of global partnerships.

In response to these shifts, and in alignment with the values and perspectives of our partners, we are embracing a new identity. The GRP Alliance reflects our continued commitment to fostering equitable, inclusive, and impactful research collaborations across the globe.

We look forward to continuing our work together under this renewed name.

We are currently revising the KFPE Guide for Transboundary Research Partnerships to ensure it remains relevant and effective in today’s evolving research landscape. This update responds to emerging global challenges, a stronger focus on equity and power dynamics, and the growing call to decolonise research practices.

Between April and December 2024, we carried out a participatory process that included 12 workshops (both in-person and online), an online survey, and expert interviews with a wide range of relevant stakeholders. A key element of the revision was a report analysing the guide through a Global South lens, developed in collaboration with the Centre for Development and Environment (CDE) at the University of Bern.

The revised guide will be structured around six principles and presented in a dual format: a concise overview document and a more detailed website offering tools, case studies, and implementation support. We are now in the consolidation phase, with the final version scheduled for publication in Autumn 2025.

We warmly invite all interested parties to stay engaged and support the dissemination of the updated guide.