December 5, 2019 Last updated December 5th, 2019 368 Reads share

The Complicated Relationship Between AI and Climate Action

Complicated Relationship Between AI Climate ActionImage Credit: DepositPhotos

There’s no other way to put it: climate change is the biggest challenge we are facing right now. In fact, it might be the greatest threat mankind has ever had hanging above its head. Climate change is such a massive and complex issue that is compelling everyone to make profound changes and come up with solutions – and fast.

Chances are you already know the dire state of the world we’re living in. The rising amount of greenhouse gases in our atmosphere is contributing to extreme weather conditions, the melting of ice caskets, and the extinction of several species. Burning fossil fuels, polluting the oceans, deforesting extensive green areas, and mass-producing meat for human consumption are all increasing the number of emissions.

All of those things are an integral part of our shared human experience, which is why it is so difficult to tackle climate change: because we’ll have to sacrifice our modern way of living. It’ll take the common effort of everyone, from presidents to farmers, from Internet providers to software testing outsourcing companies, to make the necessary shift.

In such a context, there are some people that are proposing we do so with the help of a tech ally: artificial intelligence (AI). It makes sense. AI can help us better understand the data we already have about climate change and aid us in getting more valuable insights from said data. AI-based systems could improve our processes, making them more efficient, especially in their resource consumption. It can even find new ways and technologies that could contribute to curve the greenhouse gas emissions, from new vehicles to lab-grown food.

However, as amazing as that sounds, we have to tread lightly when it comes to AI and climate action. Though a lot of experts and researchers are betting heavily on AI to face climate change, there are implications that we need to consider before trusting AI as if it were a silver bullet.

The case for an AI-based approach to climate change

Truly understanding everything that contributes to climate change takes time and research. In fact, scientists spent almost 40 years to begin to comprehend the base problem. Given that the time for climate action is now, AI can become a valuable ally to generate new knowledge about climate change. This, in turn, could lead us to the design of better strategies and stronger policies to reduce emissions and find alternatives for the staples of our way of living.

Artificial intelligence has made great progress in the last years, which means there are more sophisticated models that can gather data, analyze it, understand the causes for complex climate-related events, and recommend paths of action. It’s true that AI tools aren’t as powerful as some people like to believe, but their help in the data analysis stage can come very handy to see things more clearly in less time.

A recent paper called “Tackling Climate Change with Machine Learning” points towards the potential uses of AI to overcome climate change. One of its authors, David Rolnick, from the University of Pennsylvania, said that “it’s surprising how many problems machine learning can meaningfully contribute to.” The paper itself explains how AI can help to achieve better climate predictions, show the effects of extreme weather, and measure where the carbon emissions are coming from.

Private companies are playing their part too. There’s DeepMind, which helped Google reduce the amount of energy they use to cool data centers by 35% and whose algorithms could be used for energy efficiency elsewhere. There’s also SilviaTierra, which uses AI combined with satellite images to closely monitor the health of forest trees. Even IBM is contributing through its Green Horizon Project, which has AI creating advanced weather and pollution forecasts that have already proven to be efficient in Beijing, China.

Tech companies are the ones that are the most worried about showing their interest in tackling climate change. That could be inferred from the Tech Workers Coalition, a group of 12 enterprises that were brought together in their desire to fight against climate change. The Coalition, which includes giants like Amazon, Facebook, and Microsoft, believes that tech businesses have to adhere to a set of principles:

  • Zero carbon emissions by 2030
  • Zero contracts with fossil fuel companies
  • Zero funding of climate denial lobbyists
  • Zero harm to climate refugees and frontline communities

It’s an interesting proposal that we can only hope moves other companies to action. Yet, taking a deeper look behind the curtains shows us a couple of problems that speak volumes of the complicated relationship between AI, the companies developing AI solutions, and climate change.

The price of AI computation

What those companies don’t like to acknowledge is how much they are contributing to the same problem they are claiming to be fighting against. According to a 2018 research, tech companies will contribute 3 – 3.6% of global emissions by 2020, doubling the contribution they made in 2007, and reaching an emissions level larger than that of Japan.

That’s the invisible price we all pay for our increasingly cloud-based services and computational power. To make matters worse, running the data centers and the network infrastructure needed for platforms and tech tools implies a major use of fossil fuels. Assuming that the use of digital platforms will keep increasing thanks to the growth of cloud computing, 5G devices, and overall Internet traffic, the carbon footprint of tech companies could increase to 14% by 2040.

Want to guess what needs a lot of computational power? Yep, it’s AI. And given that the current paradigm in the artificial intelligence domain is that the more processing power an AI algorithm has at its disposal, the better results it’ll provide, chances are that the proposed AI solutions to tackle climate change can end up adding to the problem rather than solving it. That’s because, as AI needs more computational power, its carbon footprint increases.

There’s one extra thing that’s also pretty discouraging. Given that the ones investing the most in AI solutions for climate action are privately held companies, there’s no transparency to their motivations, their strategies and, ultimately, to their data. That’s why is so troubling to learn that companies like Amazon and Microsoft are aggressively marketing their AI solutions to oil and gas companies to optimize and accelerate resource extraction.

Ironically, two of the companies that are pushing for climate action from the Tech Workers Coalition are also selling technology to help the companies that are investing in campaigns against climate legislation and even promoting climate change denial.

Should we use AI for climate action?

You’re not going to love the answer: it depends. There are certain factors to control before we fully launch an AI-based strategy to tackle climate change. First and foremost, we have to make climate action a matter of international interest. Countries from all around the world have to come together to foster research surrounding specific actions against climate change, including AI strategies.

For that to happen, governments have to legislate accordingly, providing a legal framework that brings transparency to the interests surrounding investments around climate action, and enabling new green policies for the development of a new way of living.

Additionally, AI researchers have to keep working on their AI models to bring them to a new level of efficiency. That might even mean that they’ll have to change the “bigger is better” paradigm that’s turning AI algorithms into energy vampires. Making smarter technology could need further research but also a conscious decision to go in that direction.

Conclusion

Finally, we all have to realize that AI isn’t a holy savior. While it can contribute a great deal to the fight against climate change (provided that the conditions described above are met), it can’t do everything. There’s a whole other set of things we need to do as a society, such as accepting climate change, implementing environmentally friendly practices, and making the effort to change. It’s not that we have many options or time left to do it.

Climate Change – DepositPhotos

Santiago Alonso

Santiago Alonso

Read Full Bio