The Covid-19 pandemic has spurred radical measures worldwide to stem the spread of the virus, and specialists from different fields have taken up the challenge, using new technologies such as Artificial Intelligence (AI). Similar momentum and coordination also could supercharge efforts in the equally challenging drive against climate change.
The Covid-19 pandemic has had mixed effects on the environment and for climate change. On the one hand, less travel and less consumer and industrial consumption have lowered emissions and pollution. On the other hand, hard-won policy victories from before the viral outbreak such as steps to protect the Amazon and establish carbon-trading mechanisms have been put on hold. This period also has been a hard time for environmental communication and advocacy, with people dedicating less attention to climate change.
Nonetheless, there is a hopeful factor for the environment: the speed and responsiveness of the measures taken and the innovation showcased during the pandemic demonstrates that it is possible to take urgent, decisive action in the face of the deadly threats from climate change, if it were perceived to be as big and pressing a threat to humanity as the Covid-19 pandemic.
AI approaches connect Covid-19 and climate change
A recent article I co-authored about how AI is being used in the fight against Covid-19 stressed the importance of having a variety of projects and initiatives at different scales -- from atoms to societies -- in order to tackle a pandemic that is truly bigger than all of us. We highlighted the importance of sharing data and fostering projects that are led by multidisciplinary, international teams to yield fundamental breakthroughs and, most importantly, achieve on-the-ground deployment in communities worldwide. In fact, we see that the fight against the pandemic and the one against climate change are similar on many levels, not only because both are threats to the future of humanity and our way of living, but also because both require a multitude of small solutions, deployed globally, mobilizing everyone from individuals to corporations and governments. And both can benefit from Artificial Intelligence techniques at different parts of the process, at different scales with different approaches that, together, can make an impact.
Starting from the ground up, many important challenges need to be overcome at the most microscopic level to address both the novel coronavirus and climate change. For the former, this involves things like discovering molecules for vaccines and antivirals that can help neutralize the virus; for the latter, it means designing new materials for batteries, systems for carbon capture and climate-friendly chemicals. Interestingly, similar AI techniques can be used for both of these, since something that AI is particularly good at is searching through large spaces of molecules and materials to recommend which are most likely to be successful, using techniques like graph neural networks, active learning, and Bayesian optimization to learn from existing experiments and to propose novel candidates. This can help reduce the time and money needed to develop and deploy new vaccines and materials, and to explore new, innovative solutions guided by AI.
Images play a key role
Zooming out a bit to consider the type of data that AI can use in both challenges, it’s clear that images can play a key role in both. In fact, computer vision has been paramount in the fight against the pandemic, whether it involves identifying at-risk patients based on X-Rays of their lungs or their temperature or allocating hospital resources based on clinical data updated and graphed in real-time. Images also have a central role in the fight climate change, since much of the data available on the state of the planet comes from remote sensing, which involves scanning the Earth using satellites or aircraft in order to gather information, be it images or radar readings.
This breadth of information provides endless opportunities to use AI -- for example, to detect objects such as methane leaks, to monitor changes in ecosystems and track biodiversity, and even to predict the possibility of natural disasters such as floods and storms. To top it all off, these kinds of AI models are language- and country-agnostic, meaning that they can be shared worldwide to enable knowledge transfer and to give global stakeholders the ability to analyze local data.
Finally, taking another step back both Covid-19 and climate change have many similarities in terms of global impact. Given the novelty of both the pandemic and global warming, and the complexity of the systems involved, AI is a key tool in modeling and predicting the impacts of these two phenomena. Modeling the spread of the coronavirus using AI involves using different data sources, from social media posts and Internet searches to counts of infected individuals published by the WHO and national agencies. Modeling the extent of climate change and predicting its future impacts also involves a cornucopia of data sources, ranging from statistical models to represent the physics of clouds and improve weather forecasts to audio recordings of bird song to track biodiversity.
Given enough data points, AI techniques can help governments make the right decisions at the right time, since combining data and knowledge-driven modeling makes it possible to better estimate future consequences of present actions, be it the extent to which school closures and social distancing will curb the spread of the virus, or the impact that global warming will have on the melting of ice caps and the rising of sea levels.
Sadly, not everything is rosy when it comes to applying technologies like AI in complex, global endeavors like the pandemic and climate change. In fact, both face similar challenges. For example, it is crucial for AI researchers to collaborate with experts in the field of application that they are targeting. For example, for actions at the atomic level, this can mean working with chemists and biologists, who not only have the necessary domain expertise but also access to infrastructure that can synthesize and evaluate candidate molecules and materials. This kind of collaboration isn’t always easy, since it means that everyone involved must learn new concepts and different ways of seeing familiar phenomena, and to leave their comfort zone and expand their horizons and ways of engaging with their work.
Converting breakthroughs into real-world applications
Above and beyond pure research, there are also many challenges involved in translating fundamental breakthroughs to real-world impacts. This can only be ensured by building meaningful partnerships between the public and private sectors, and between researchers and entrepreneurs or non-profit organizations. Only this enables taking fundamental solutions from the pages of scientific journals and into the real world. This is a challenge in and of itself, since it involves taking into account legal and operational aspects of deployment, as well as budgetary constraints and the messiness of real-world data and applications.
AI techniques also are only as good as the data being analyzed. So the key to developing realistic models to guide collective decision-making remains gathering quality data on different levels and taking into account the multitude of potentially relevant factors that are involved.
The world is currently in a time of crisis, and that also offers opportunity. Governments are injecting tens of trillions of dollars into the global economy to sustain economic activity, and this capital has the opportunity to determine the carbon intensity of our economy in the next decade. This provides a chance to invest this money into multidimensional technological solutions, ones that will improve health, energy efficiency, urban infrastructure, and sustainability. AI is a key piece of that puzzle, and an enabler for a global strategy to build a resilient, sustainable future.
Many open questions still remain, from quantifying and reducing the carbon footprint of large-scale AI models to incorporating considerations of equity and ethics into the design of AI algorithms. But these issues are gathering momentum and becoming increasingly paramount in AI research and practice.
Addressing climate change requires rapid, sustained, and scientifically informed efforts involving a diverse set of stakeholders. Both the Covid-19 pandemic and climate change present similar global challenges that transcend borders and continents while emphasizing the value of global well-being and partnership. And efforts to address both can benefit from similar AI approaches to deploy actionable and meaningful solutions.
AI has finally left the research lab and is being used in real-life settings to generate positive impact. While Covid-19 is undoubtedly the most imminent global threat, climate change is not slowing down, but simply gathering momentum while our attention is elsewhere. There are hundreds of potential AI approaches for climate change that can be used by all levels of society to pave a sustainable post-Covid future. It is up to us as individuals and as a society to make them a reality.