Public Hearing on AI and the Green Deal
After some technical difficulties (ironic, I know), Victor Galaz managed to share some prepared reflections at the latest public hearing “AI and the Green Deal” organized by the European Parliament and its Special Committee on Artificial Intelligence in the Digital Age (AIDA). Here is a written version of his reflections:
Dear colleagues and members of the Special Committee.
First of all, I would like to thank the Chair and organizers of this hearing for inviting me to speak to you today on this important issue. Since my time is very limited, I would like to focus my contribution on 3 observations, 3 concerns, and 3 recommendations that I believe are key as we all strive to make sense of how planetary change, digitalization and AI are - and should be - related. Let me start with the observations.
The first observation is that the definition of “climate” whenever climate change and AI is discussed together, often is too narrow. Applications of AI span beyond climate modelling and prediction, and beyond optimization of energy systems and traffic flows. The climate system is fundamentally connected to our living planet; biodiversity, forests, the oceans, agricultural ecosystems, and more. Embracing this connection with the living planet, and our role in it, is key if we are to promote a constructive agenda that brings together the responsible development and deployment of AI to tackle urgent climate and sustainability challenges.
My second observation is that applications of AI include issues that are broader than climate mitigation and optimization of resource flows. AI also has the potential - if used responsibly - to help people, communities, civil society, governments and the private sector - to reduce climate risks, to support climate adaptation, and at best even inspire climate innovation and transformation in new ways. Such a broader view is urgently needed today.
My third observation is that the time to direct and steer the full potential of AI for the climate challenge, is now. There is a growing interest in society about these issues, and both the private sector and research community are mobilizing and shaping their agendas as we speak. And needless to say, there is an urgency to accelerate climate actions that help bend emission curves and support climate adaptation and innovation in society as a whole.
So in sum, the potential of AI for the climate challenge is not mainly in optimizing systems, but in augmenting the capacities of people to address the climate challenge in all its dimensions. I also have 3 concerns. The first is related to the risk of Hype. As the pressures on our living planet and the climate system increases, so does our hope that we will be able to find AI-solutions to deeply complex and challenging social, economic and environmental challenges. Our knowledge about whether AI actually does offer large climate benefits is highly limited, and all assessments that have been made so far are preliminary, and at times I would even say, too optimistic. This hype is to a large extent driven by the private sector, and could use some solid reality check as the field evolves over time.
This connects to my second concern, Acceleration. The deployment of AI-systems and related technologies like IoT, 5G and robotics may very well lead to accelerated extraction of natural capital and loss of ecosystems. I’m not only thinking about the continued extraction of fossil fuels, but also increased expansion of monocultures on land- and seascapes, loss of biodiversity, and of course an increased pressure on the raw materials that underpin these technologies such as rare minerals. If we are to use AI in a planetary responsible manner, we need to find ways to break this cycle of acceleration of business-as-usual, and the loss of resilience that has created the problems we are facing today.
My third concern is Lack of Transparency. All of us investigating the real-world applications of AI-systems discover very quickly that access to information about models, data and social-ecological implications is treated as proprietary by the private sector. This is concerning because AI-innovations for sustainability is dominated by the private sector, which drastically reduces the possibilities for us to investigate these technologies and their impacts in the world. As we all know, limited transparency means limited accountability. We should not accept this. So, how can and should the EU and representatives in the EU Parliament try to address these challenges, and support the planetary responsible development and deployment of AI?
Three aspects are as I see it, key:
The EU needs a clearer agenda on how AI could help support planetary stewardship and resilience. It should integrate our living planet as a key aspect of the climate system, and recognize that AI could support not only mitigation, but also climate adaptation and innovation. The world doesn’t need more AI, it needs smarter AI.
The EU needs to develop better overview capacities as AI-augmented technologies move into domains critical for climate stability and a safe and just transition towards sustainability. That means monitoring, supporting and - when needed - even challenging the deployment of AI-technologies as they shape our ecosystems, the climate and even our perceptions about our changing planet.
The EU needs to explore and use its influence as a large investor in these technologies. This means demanding that companies and government agencies make sure that their data is “findable, accessible, interoperable and reusable (FAIR) while ensuring the fairness, accuracy, confidentiality and transparency (FACT) of the algorithms and tools they create.”
I realize these are all big questions and big challenges. But again, the time to discuss and direct the full potential and force of AI for the climate challenge and resilience, is now. Thank you for listening, and for your engagement on these issues. I look forward to our continued discussions.