3.02 AI in environmental projects

A spectrum of applications

After establishing the potential for synergy between AI and environmental knowledge, it is crucial to examine the practical applications and real-world challenges of using AI. The technology offers a broad spectrum of tools that can be applied to environmental action. These applications generally fall into three main categories: monitoring, prediction, and optimisation.

• Monitoring: AI excels at analysing vast amounts of data from remote sources like satellites, drones, and sensor networks. This allows for large-scale monitoring of environmental issues such as deforestation, air and water pollution, and changes in biodiversity. By processing this data in near real-time, AI can help detect problems like illegal logging or chemical spills much faster than traditional methods.
• Prediction: By analysing historical and real-time data, AI models can forecast future environmental trends and events. This includes modelling the long-term impacts of climate change, predicting the path and intensity of natural disasters like hurricanes and wildfires, and anticipating periods of high flood risk. This predictive capability enables a shift from reactive responses to proactive preparation.
• Optimisation: AI can be used to make existing systems more efficient and sustainable. In agriculture, algorithms can help optimise irrigation and fertiliser use to maximise crop yields while minimising environmental impact. In the energy sector, AI can optimise the output of wind and solar farms and improve the management of smart grids.

A critical perspective on AI solutions

While the potential applications are vast, it is important to approach AI with a critical perspective. AI is a powerful tool, but it is not a universal solution, and its application comes with significant challenges.

One common pitfall is the “when you have a hammer, everything looks like a nail” syndrome, where complex AI solutions are proposed for problems that might be solved with simpler, more direct methods. There is a risk of creating over-engineered systems when a more straightforward approach would be more effective and less resource-intensive.

Furthermore, the effectiveness of any AI model is fundamentally dependent on the quality of the data it is trained on. AI systems can perpetuate and even amplify biases present in their training data and using incomplete or poor-quality datasets can lead to inaccurate or misleading results. This highlights a critical need for high-quality, reliable, and contextually relevant data, which sets the stage for the importance of community-driven data collection efforts like citizen science.

Finally, it is important to consider whether an AI solution is addressing the root cause of a problem or merely its symptoms. For example, while AI can be incredibly effective at monitoring and tracking plastic waste in the ocean, it does not address the underlying issues of overconsumption and inadequate waste management that create the problem in the first place.

Ultimately, deciding whether AI is “good or bad” for the environment is like trying to decide whether using rockets is good or bad by looking at a space shuttle and a missile. The technology itself is neutral; its impact depends entirely on how it is applied. The key is to thoughtfully select the elements of AI that are helpful and leave those that are not, ensuring that the technology serves as a targeted tool to solve specific, well-defined problems rather than being seen as a panacea for all environmental challenges.