AI in the Wild book cover. Credits: AI in the Wild

Sorry to burst the bubble, but “artificial intelligence is never going to produce a sustainability revolution.” At least not according to Peter Dauvergne, author of AI in the Wild: Sustainability in the Age of Artificial Intelligence.

AI in the Wild is impressive when considering that it is the first book on its subject. Coming from a background of research in technology, ecosystem degradation, consumption, social movement politics, social inequality, and corporations, Dauvergne is a skilled analyst. However, the book follows many topics, which can make it hard to follow as it jumps around from conservation to smart cars to weapons of destruction. AI in the Wild’s biggest strength is the examination of the potential pitfalls of AI.

AI in the Wild presents many examples of the “complex, hidden, and capricious ways” that AI can harm the environment. Some of the most thought-provoking examples include:

• The use of AI by fossil fuel companies could add hundreds of billions of dollars to the oil and gas industry annually by helping to find new fossil fuel sources, reduce labor costs, and ramp up production.
• The use of AI in social media, search engines, and store cameras has the potential to increase the global retail industry by hundreds of billions, if not trillions, of dollars per year, thus drastically increasing greenhouse gas emissions.
• The significant amount of data needed for systems like AI will result in data centers being responsible for more than 7% of global greenhouse emissions by 2040, according to the Journal of Cleaner Production.

These examples support the bigger picture Dauvergne paints. Though AI may be very advanced technology, it still cannot solve underlying issues of the environmental crisis such as resource politics, capitalism, and inequality.

In regard to inequality, AI in the Wild reminds readers that it is important to consider the impacts of AI on fragile ecosystems and marginalized people. While Dauvergne’s argument isn’t necessarily specific to AI, he does note that AI will disproportionately benefit those who can afford it. North America and China in particular are projected to hold “70 percent of the economic value of artificial intelligence over the next decade”. Marginalized communities that do not have the financial means to invest in AI will not see the same benefits. 

Plus, any environmental consequences, such as increased greenhouse gas emissions, of AI will leave even more environmental degradation for marginalized communities. These communities are already more vulnerable to climate issues such as sea level rise, drought, and increased heat in urban areas. So, while marginalized communities aren’t the primary ones causing the harm, they are certainly on the receiving end of it.

And the activists taking a stand against it? AI is putting them in even more danger than before. Environmental activists have been repressed globally for decades, with particularly severe risks in ‘developing nations’. Dauvergne notes Honduras specifically as a country where more than 125 environmental activists were murdered from 2010 to 2020. As Dauvergne acknowledges, “since at least the 1960s, security forces have been spying on activists, infiltrating grassroots movements, and instigating violence”. AI surveillance in the hands of bad actors only increases the risks of being an environmental activist.

Thankfully, Dauvergne notes that the world isn’t full of bad actors. Consider the researchers who created LarvalBot. LarvalBot is an underwater robot using computer vision AI to help restore Australia’s Great Barrier Reef. LarvalBot delivers coral larvae to areas of the reef that have been harmed by pollution, cyclones, predatory starfish, and climate change. This careful use of AI is making a positive change for the environment.

The main takeaway of AI in the Wild is that because there are valid environmental concerns to artificial intelligence, responsibility is key. Unfortunately, as with other tools, AI can’t be kept out of the hands of bad actors and it can’t solve every problem, such as consumer culture. Dauvergne states that to use AI responsibly, what’s needed is “diverse, cautious, and self-aware teams of programmers” and “more jurisdictions to regulate artificial intelligence, the internet, and big data collection”. So, while AI may not be producing a sustainability revolution any time soon, learning to responsibly utilize AI as a tool for sustainability can be an asset towards a better future.