Lessons from a trip to the Ars Electronica Center in Linz, Austria
Digital Europe
Far from the talk about the ethics of Artificial Intelligence (AI) and if it will be used for good, is the question of the inequality in AI development for and by the data-scarce regions of the global south, and how these regions are precluded from truly enjoying the fruits of AI development on par with other regions of the world.
The world’s largest and most disruptive AI models are trained with unimaginable amounts of data. When the pools are saturated with data from certain regions of the world and not others, an inherent and inevitable inequality martializes. Several studies have identified such inequalities including the Computer Vision Fails exhibition at Ars Electronica.
As a person who subscribes, not only the UN-GGIM mandate to bridge the geospatial data divide, but also the digital earth principle, and having experienced life in a part of the world on the verge of creating a fully-fledged digital twin of the real world, I hear of statements such as open data, tax-funded data, data for development, accessing the most sophisticated commercial data (e.g. very expensive satellite data) for research purposes, seemingly for the first time. Over this past year, I have come to learn that the technical and scientific research and development evident in this new world, is underpinned by data. FAIR-compliant, free, open, and available data fuels innovation and development.
Learning about the EU’s INSPIRE Directive through the Copernicus Master's in Digital Earth (CDE) OpenGIS Standards, Architectures and Services course, and getting to know the origins and offerings of SAGIS through the CIVIS BIP Participatory tools for Urban Nature Planing and Management Summer School, has helped me understand the importance placed on open government data and the pertinent role it plays in the region’s technical and scientific development.
Data Scarce Global South
The same cannot be said for these so-called data-scare regions. Coming from one such region, where everything is largely analog, with most public data collection and storage services being pen and paper (or even typewriter) based, I have witnessed firsthand the inefficiencies that plague such a system. It explains why some of such places are called data deserts where data scarcity is the norm, data is either unavailable or very expensive. This is not to say that there are no national statistical bodies, but when they operate archaic 19th-century methods in this 21st-century digital age, the widening of the so-called digital divide is inevitable.
When public data collection is done, which comply with modern standards and conventions and utilize public funds, but are not made publicly available, we are indeed talking about the same thing. It is shocking to see evidence of valuable geospatial and remote sensing datasets, such as very high-resolution orthophotos and Lidar point clouds of entire regions in Nigeria, created and stored but made inaccessible to the public (even for research purposes). Yet, Nigeria lacks adequate DSMs and locally produced and validated high-resolution DEMs that could prove invaluable for disaster risk prediction and management, especially for chronically occurring phenomena like flooding.
Granted that most of the fastest-growing, disruptive AI platforms (e.g., OpenAI) are proprietary, the role of open data in the research and development, done in the build-up to these magnificent platforms cannot be under-emphasized. Therefore, if we hope to attain similar levels of scientific and technical success back home in the data-scare regions of the world, we truly must emulate some of these principles, while considering peculiarities of course.
Food for Thought
I solely believe that when these have been put in place, the work of analyzing problems, proffering solutions, abstracting, automating, and analyzing these solutions becomes much easier. Imagine training a deep neural network, the average performance model needs thousands of data samples to learn from.
It begs the question - When data is scarce, how can AI thrive and be used to solve societal problems?
While you ponder on that, enjoy below a short clip summarising my amazing experience at the Ars Electronica Center visited as part of the requirements in the course, Actors in Corpenicus, of the CDE Curriculum.
Cover photo courtsey Christian Holzinger on Unsplash