Fighting Against Climate Change “The Question Remains: Are We too Slow?"
We are in the midst of climate change and must use AI to tackle this challenge, says deputy director of the Stockholm Resilience Centre Victor Galaz. At the same time, AI technologies which harm the climate further are being used. How can we find a balance between these poles?
By Svenja Hoffmann and Natascha HolsteinMr. Galaz, you are the deputy director and associate professor at the Stockholm Resilience Centre. When in your professional career did you notice that AI can have an impact on climate change?
In the mid-2000s, I became increasingly interested in how the World Health Organisation (WHO) in Geneva used unofficial online sources of information to get early warnings of epidemic disease outbreaks, for example, online reports of sudden increase of pain medication sales in Asia. The official channel of the WHO had previously been to wait for country reports informing them about disease outbreaks. The new system, which was officially in a grey zone for some time, used machine learning to explore online news data and extract early warning signals. This development triggered my interest of how information, its analysis and technologies could transform organisations, legal frameworks and perceptions and it has followed me over a very long time.
All along, my area of interest has been the challenge and urgency likewise to tackle climate and sustainability issues. With the continuous development of AI and deep learning algorithms, technical changes in robotics and sensors, the accumulation of data – we need to question their potential and risks at the same time.
You just mentioned potential of AI. How can AI be used to fight climate change?
We can use AI for helping people to adapt to climate variations, such as farmers, all the way to big banks, to prepare for climate change and the impacts it may have. We are already experiencing a changing climate, so we need to use AI to make sense of it and to adapt in a proactive way to reduce risks. There are many AI-based methods to minimize the use of resources, of energy and emissions. Scientists are also becoming excited about the potential of AI to get improved data and predictions of changes on the planet. For example, to be able to map and understand extreme weather events in a different way – sea level rise, changes in marine systems. AI is an excellent tool to get a better view of what's happening on the planet.
AI is an excellent tool to get a better view of what's happening on the planet.
At that time, when the interest around AI accelerated globally, it related to “Terminator”- framings, with associations of AI becoming self-conscious and destroying humanity. The Biosphere Code was meant for companies, governments and the public – basically anyone interested in AI innovation that is responsible for the planet to get inspiration and guidance. We had people doing financial hacking, artists, game developers, philosophers, a few academics.
The Biosphere Code didn't have that much impact, but that wasn’t its ambition either: rather than a fixed protocol for companies, it was meant to be the beginning of a conversation to reflect on the present implications of technological development and its impacts on our planet and people. There is a gigantic gap between people developing AI technologies and those developing sustainability and focussing on its issues. There has been considerable progress in the last years concerning AI’s ethical and social implications, but in terms of sustainability and planetary stewardship, the development has been much slower.
As an example, many farmers worldwide have been hit very hard by climate change within the last decade. How could farmers in rural areas, for example in the so-called global South, make use of AI technology?
There is an impressive group of people working on big data in agriculture, for example, the Consultative Group for International Agricultural Research. They help develop tools which give small farmers guidance on how they should farm, such as information to predict market development or the weather. These simple applications are offered for free and are accessible on the farmers’ mobile phones. We may see more of them, especially if they can be linked to insurance solutions. For instance, after a heat-sensor has been installed and the heat passes a certain threshold, farmers can get an immediate payout from the insurance company. This potential lies also in coastal fisheries or city planning.
However, there is a big digital divide. Farming communities that have very few hectares don't have excess to advanced technologies. You may have a mobile phone, but maybe not 4G. You may also not have an autonomous tractor or a drone. Many technologies benefit from big scales, like having a substantial company or large agricultural areas. This poses a risk for small communities because the powerful technologies will not be accessible for them but only for big players. It is crucial to bridge this digital divide.
There is a big digital divide in the farming industry: the most powerful technologies are accessible for big players, but not for small farmers.
That's a difficult question. We cannot assume that the private sector will solve this problem without strong incentives. However, there is much interest in targeting and supporting the most vulnerable from the beginning with the investments in climate and digital innovation, whether they are from countries or multilateral organisations like the World Bank. This is an opportunity.
Does the rapid progression of climate change influence the development of “good” AI to tackle those challenges or is it the other way round?
There is growing interest from prominent organizations such as the World Economic Forum, Microsoft and others, in applying AI to cope with urgent climate challenge and develop the technology further. In my opinion, this is positive because we need everything that is available to us to solve the climate crisis. At the same time, many of these technologies are already being used to accelerate climate change. Deep learning algorithms and big data analysis have or are still being developed to find new sources of fossil fuels with the aim to sell the service to fossil fuel companies.
Since the technology to discover and extract fossil fuels has already been created, isn't it too late to install guidelines to stop the progression of these actions?
Technology nowadays allows you to go further than you could before. For example, you can extract minerals from the ocean bottom which you couldn't do five or ten years ago. On the other hand, coal is slowly dying as a source of energy and the same accounts for other natural resources. So, yes, technological development will continue, but in the end, economics and their political limits will decide over its use.
Even though it's technically possible, from a long-term perspective, it doesn't make economic sense to pump up fossil fuel.
How can we give huge companies economic incentives to use AI in a responsible way? Don’t they need to do the first step?
I wouldn't say companies need to move first and afterwards governments and finally consumers. That is not how change happens, they all need to move. We need legal boundaries that are decided upon by voters, their governments and potentially international regulations. For instance, in the EU, we have the GDPR, regulations that were agreed upon by governments on how to manage data about individuals.
Also, consumers can pressure enterprises – because consumer decisions matter a lot. They can object to certain products because the company is using data in an unethical way. Just like many people called out companies that use palm oil or sell beef from Brazil. The same accounts for investors: Big tech giants have owners who can use their influence to stir the company into this direction or another. In conclusion, we need a combination of all of them.
We need legal boundaries, pressure from consumers and investors to make tech-giants act ethically and sustainably.
In your “Couch Lesson,” you said we need planetary responsible AI. What do you mean by that?
It’s called responsible AI because it is about transparency, accountability, making sure that algorithmic or AI systems don't have discriminatory impacts. There is growing recognition for it, many big tech companies even have responsible AI units. But if we want AI technology to contribute to a sustainable future for all, we must also focus on climate and environmental issues. That is the notion behind planetary responsible AI which includes, first of all, looking not just at climate but all sustainability issues such as biodiversity, water, ecosystem resilience etc. Secondly, recognising that these technologies need to be developed with respect and together with people that are knowledgeable about our living planet. Thirdly, the applications need to have distributional effects which should support the most vulnerable communities in a positive way, rather than creating more risks or changes in benefits that are skewed, like mentioned earlier in digital farming.
What do you predict is going to happen in terms of the positive and negative use of AI concerning climate change?
There will be mixed uses of AI for various things, some of them will be good and some of them will be bad. And some of them will be in areas where we didn't expect AI to move into and have a big impact. Just think about how machine learning and social bots are used to diffuse misinformation on digital platforms in ways that we couldn’t predict five years ago.
What do us and governments need to do to tackle climate change, not specifically concerning AI?
First of all, we need a strong international agreement since it's a global issue. The Paris agreement has been created – now its ambition needs to be implemented. Thus, secondly, we need national action that is tangible, for example setting a net zero-emission-target. Some of the biggest countries in the world have committed to this goal: China has promised to reach carbon neutrality by 2060, the EU by 2050, we are hoping the US will follow suit with the new president. I believe other countries will line up because if they move early they might get an economic advantage. They can develop technologies and export them later. There is a lot of self-interest in it, but it is good because then changes move quickly.
At some point, we need to start pricing emissions because the economic incentives to emit CO2 are just too weak in many countries at the moment. If and how it works, will depend on the political context. Sweden started early with carbon taxing and it is one of the highest ones in Europe, so today, it is not a big political issue but just part of our reality.
Do you have hope?
Yes, I'm optimistic based on data. Some days I'm pessimistic, but as already mentioned, technological change together with economics is making fossil fuels a so-called “stranded asset” and that's exactly where you want to be: you want companies and investors, banks and governments to see that they are losing money by continuing to invest in fossil fuels. When this perspective becomes apparent, shift happens quickly, and we are in that shift right now. However, the question remains, if it is going too slow, and whether other important changes are ignored like biodiversity loss and deforestation. As a person who works in the sustainability sciences, I would have preferred the shift to start 20 years ago. The current emission problem would be much smoother, whereas now, the transition is abrupt and might be disruptive in a negative way.
Shift happens quickly and we are in a shift right now. The question remains, if it is going too slow.