Together, we can empower 100 million farmers with weather intelligence by 2030 - join us!
5 Major Takeaways From The Climate Change AI Global Summit
April 28, 20208 min read

5 Major Takeaways From The Climate Change AI Global Summit

We at were excited to participate in the very first virtual Climate Change for AI conference this past weekend. The event, originally planned to be held in Addis Ababa, Ethiopia benefited massively from the format of a virtual conference as it meant there was broad and truly global participation with 6 continents represented across 87 countries more than 2,600 attendees, including speakers from Google AI, Microsoft AI for Earth and The purpose of the event was to share impactful uses of machine learning in reducing and responding to climate change, and as an energizing virtual venue for discourse between experts across a range of stakeholders — researchers and engineers, entrepreneurs and investors, established businesses, and the public sector.  

The team was invited to share their approach on how to improve access and usage of life transforming weather information in Africa through advanced weather technologies and the role that ML and AI can play as a critical contribution to the climate adaption agenda, with Georgina presenting an opening keynote talk and Daniel,’s Chief Scientist joining for a fascinating panel discussion with Dr Ciira wa Maina from Dedan Kimathi University of Technology in Nyeri, Kenya; Paula Hidalgo-Sanchis from United Nations Global Pulse; Sarvapali Ramchurn, Professor of Artificial Intelligence. Here are a few key takeaways: 

1. Weather Data Can Save and Transform  Lives:

By improving weather products and services in Africa, we have the potential to unlock hundreds of billions of dollars of opportunity for the continent, and, in turn, lift millions from poverty by helping individuals, businesses and governments proactively manage weather risk and adapt to changing weather patterns. has a new approach to bringing innovation across the hydromet value chain to provide long term sustainable solutions for improved forecasting in emerging markets. We have founded to ensure that the massive socio-economic benefits of these advanced technologies reach the most vulnerable communities.

2. Public Private Community Partnership Matters:

To drive sustainable long term solutions, we must work together as a system to drive change and ensure that the technologies and approaches reach their full potential in Africa. Each stakeholder has a critical role to play in advancing progress, from academia, to local government to private sector. We must provide forums for open data sharing; We must ensure we are listening and learning from the end users of the information and insights, especially those who are most vulnerable, to implement solutions that matter; And we must work together to create a system that is sustainable and economically viable. is committed to finding effective ways to work with the system to drive change.

3. Why Does AI Matter to Weather and Climate?

AI and ML have massive opportunities for improving weather forecasts and business insights. For example, leverages machine learning to help hyper-localize our weather data products. This involves applications such as enhancing quality control and inferring weather information from the “weather of things” data sources we consume, as well as producing fine-tuned forecasts for specific locations and applications.

We see the biggest opportunity for AI and ML in improving and automating decision making with regards to changes in the environment. Computer vision techniques can help unlock the potential of satellite data to provide automatic, near-real-time analyses of evolving conditions for remote areas where communication may be difficult. ML-based post-processing of weather forecasts can more directly tie them to “on-the-ground” decision making, adding needed context for how the weather forecast will impact local communities. AI-based modeling approaches may yield alternatives to complex, expensive traditional ones that are often used for hydrology or other applications. Finally, as a basic pre-requisite, developing any AI/ML solution requires curating a very large body of data which itself will be extremely relevant and useful to for many stakeholders, either directly or as part of derivative spin-offs.

4. What Are the Biggest Opportunities for Climate Change AI to Use Weather Data? 

We also see MASSIVE opportunities in how the Climate Change AI community could use’s weather data. We are offering free access to our API, and encourage the community to experiment and seek to add value in their local context. We are eager to promote inclusive data reach and collaboration with local partners. If you have an impactful application of our data in the context of Africa, please do reach out on our website: Here is where we see opportunity for weather data and impact.

  • Value-added forecasts, especially for parameters of interest and impacts beyond the standard weather forecast
  • Downscaling global weather models for local information, e.g. training for specific locations where data might be available
  • Automatically identifying hazards in the forecast, to slightly reduce the need for humans “in-the-loop”. For example, it’s possible to use Natural Language Generation to write textual summaries of time series of data, say a weather forecast. These traditionally do not place weather in the context of local climate (e.g., higher than average temperatures or chances of flooding) but there is no technical reason why they couldn’t. Automatically generating these contextual forecasts and providing them directly to farmers or stakeholders could empower their decision-making processes.

5. What are the biggest challenges for AI/ML in the context of Africa?

  • Data: We need larger archives of more reliable and robust ground data to train models, and to incrementally build up ground truth
  • Data Sharing and Analysis Platforms: Data needs to be collected and made accessible, ideally by leveraging new strategies with data/compute locality on the cloud and by preparing the data in easy-to-use and understand formats with very good documentation and use case examples
  • Domain Knowledge and Context: Don’t assume that the “flavor” of problems we solve in one country are the same as those faced by stakeholder communities in another These can range from technical (different meteorology in the Sahel versus New England, with different implications for forecasting) to societal (flash flooding in remote areas with paved roads, as in the central US, is not nearly as significant a problem as when it happens in central Africa, where many communities may have poor infrastructure much more susceptible to these hazards) to political (different regulatory regimes and motivations, for instance with respect to air quality measurements)  
  • People: Disconnect between the scientific/AI  community and users on the ground/ knowledge of real use cases and applications
    • Trust: “Black box” or “crystal ball” AI solutions are inherently untrustworthy and we need to find ways to create trust and transparency in them
    • Solving the “right” problems: We need to identify specific, actionable problems with stakeholders on the ground in Africa and solve these particular challenges. We want to help people solve their specific problems.

If you’d like to learn more about the current technologies we can be implementing now to improve the lives of billions and drive real economic impact, we’d love to talk to you. Learn more about and let’s be a part of crafting a better future together.