Working Out Loud

Unfortunately, we had to take the DPPD atingi course offline due to an operating model change on the platform which made the costs unsustainable. However, we are currently working on alternative ways to make the course materials available elsewhere in an euqally interactive format. Any tips or suggestions would be very welcome!

After our return from our eWaste Computer Vision field trip to Ghana, with over 5,000 photos taken (!), we’re now busy labeling and refining our model. Labeling means tagging each image by hand so the model can “learn” and improve. It’s detailed work, but progress is underway. Stay tuned for more updates!

The year started off busy! We’re currently testing the FempowerBot prototype (still under wraps, sorry 😉, it will be a training enhancement tool that allows users to trial difficult conversations around the topic) through user interviews and just wrapped up a specifications workshop to nail down the details for the next phase. We’re grateful for the amazing experts joining us on this exciting journey. 🙏 The milestone of our specifications reference document is just around the corner!

Our e-Waste project is becoming concrete! We have contracted partners in Ghana and Germany and are looking forward to a trip to Ghana at the end of October. We plan to visit scrapyards to get a firsthand impression and collect pictures for training advanced AI image recognition. We are looking forward to meeting our local partners and to entering this exciting new phase of the experiment.

Just in time for the resumed negotiations of the UN Plastic Treaty in Geneva, the Data Lab is pleased to announce the official launch of NegotiateAI 3.0, now available on its own domain, with a fresh new design and new features. Data Lab team members Teresa Kroesen and Robin Nowok will be on the ground in Geneva for the next two weeks, collecting live impressions from what is expected to be another round of tough talks. The two will also present the new NegotiateAI application at the PREVENT Waste Alliance Side Event.

Together with GIZ’s PoMuC Project in Brazil, we have been working on integrating Data Feminism into the structure of a new project. This involves analyzing the module offerings for entry points for Data Feminism, co-designing a workshop intended to identify potentials, and commenting on the Terms of Reference for the integration of Data Feminism into the Climate Justice Strategy of the program. Further collaboration is envisioned for the future.

At the GGKP & SwitchAsia webinar “Harnessing AI to Scale the Impact of Circular Economy Knowledge”, we had the opportunity to discuss how AI can help make circular economy knowledge more accessible, structured, and practically useful. A key takeaway: The real value of AI lies not in how smoothly it generates text, but in how reliably it connects to verifiable knowledge. To be useful in policy and practice, AI must be grounded in curated sources and offer full transparency – users need to trace the origin of its answers.

One idea to connect NegotiateAI’s “plastic knowledge hub” to an API integration with plastic taxonomies of the GGKP to link plastic-related knowledge and data models, helping users navigate complex sustainability content with precision.

The recording is available here: https://www.switch-asia.eu/event/harnessing-ai-to-scale-the-impact-of-circular-economy-knowledge/link

GIZ’s Project “Feminism in Action for Structural Transformation (FAST)” will be launching a Community of Practice on Data Feminism, hosted by Ladysmith Collective. We at GIZ Data Lab are thrilled to be a part of it!

In the aftermath of INC5.1, we have carefully analyzed the extensive feedback and improvement opportunities, setting the stage for the next phase of our application. Huge steps are ahead: We are working towards an independent platform featuring analytical dashboards, more performant models, real-time data integration, and much more. Scaling up to this next level will take several months, but we are excited about what’s to come. A key priority is the early involvement of partners from EL, SIDS, and other resource-constrained delegations in the plastic negotiations. We’re looking forward to co-developing solutions that truly meet their needs.

Together with our colleagues from the German Data Labs of BMZ and Foreign Ministry, we are exploring ways to scale elements of NegotiateAI towards a jointly developed negotiation AI tool for climate action. Over the past weeks, we’ve been consulting with climate negotiation experts or people who actively participate in climate conferences and bring deep expertise in both climate policy and negotiation processes. Our goal is to pinpoint concrete use cases where AI can support and enhance negotiation efforts. Next step: prototyping based on the insights gathered, we will start testing AI-driven solutions for strategic insights, scenario modeling, and decision support.

We had the pleasure of giving a keynote at “AI for a Greener Future: Innovations in Environmental Protection”, hosted by the PlasticObs+ consortium. Together with researchers, policymakers, and practitioners, we explored the dual role of AI in supporting environmental goals: its potential and its pitfalls. In the talk, we highlighted the example of NegotiateAI for structuring and navigating complex knowledge in international plastics policy. The discussion made clear: AI can be a powerful enabler, but only if it’s used transparently, with clear sourcing and context-aware design. It was encouraging to see so many actors from science and practice critically engaging with the topic. Thanks to PlasticObs+ for creating such a thoughtful space for exchange.

Together with the Sector Programme Feminist Development Policy & Gender Equality, GIZ Data Lab is working on a Chatbot to help make the results of a GIZ-internal toolkit on Feminist Development Policy more tangible. By using the Chatbot, colleagues shall be able to train their knowledge on the content of the toolkit.