With our world becoming more dynamic and connected than ever, new technology is revolutionizing how we study the environment at an incredible pace. One of the researchers leading this transformation is Professor Tal Svoray, a geoinformatics expert at Ben-Gurion University of the Negev.
His research centers on developing AI methodologies for detecting and analyzing environmental phenomena using aerial imagery. “Two years ago, we were caught by the huge tsunami of AI,” Prof. Svoray explained in an interview. “Deep learning can provide a much more educated space-time analysis and identify objects far better than traditional statistical tools.”
He and his team harness artificial intelligence (AI) and space-time data to tackle noteworthy environmental challenges, from monitoring soil erosion to predicting tree mortality. Their work pushes the boundaries of how we understand and protect our planet, using cutting-edge technology to make sense of complex ecological changes.
Prof. Svoray’s enthusiasm for AI stems from its unprecedented capacity for pattern recognition and predictive accuracy. Traditional physically based models require extensive manual calibration, while AI-driven models learn autonomously from vast datasets. “Machines can analyze entities on the Earth in a much more structured way than statistical tools developed by humans,” he said.
As head of the Geographic Information Laboratory (GI-Lab), Prof. Svoray collaborates with a dedicated team of researchers and students, including Roni Blushtein-Livnon, Osher Rafaeli, Idit Tikotzki, Amir Boger, Keren Sandberg Ashkenazi, David Ioffe, and Einat Bela Liron.
He and his team at GI-Lab have implemented AI-based techniques to track solar panel deployment among Bedouin families in the Northern Negev, map sinkholes and soil erosion in agricultural regions, and assess the impacts of climate change on tree mortality.
Hidden within these titles is crucial information contributing to efforts such as supporting the transition to renewable energy in marginalized communities, promoting sustainable development, understanding land degradation, which informs better land management practices essential for food security, and developing effective conservation strategies. This shift has allowed his team to extract crucial insights into climate resilience and sustainability. They use AI to predict flood risks and model long-term environmental patterns.
The ability to automate the detection of environmental changes has transformed the field, making it more efficient and precise. For example, his team applies AI to study tree mortality by analyzing photosynthesis activity over time. “We study the photosynthesis of trees by inputting satellite data shot every 15 days since the 1990s. This cannot be done using regular statistical tools,” he emphasized.
The ability of AI to process decades' worth of satellite data has transformed how researchers approach ecosystem monitoring. “Deep learning allows us to extract patterns and analyze vast datasets that would take humans years to process,” he explained.
Integrating AI into geoinformatics is not without its challenges. The transition from physically based models to deep learning requires processing enormous datasets. “AI can be a ‘black box,’ meaning that, while it produces results, the reasoning behind its decisions is often unclear,” he admitted.
However, his lab has explored the performance of human cognition against machines by comparing human annotators with AI analysis of aerial photography. “In several cases,” he revealed, “AI outperformed humans who had been doing this daily for two years.”
Another key challenge is dealing with imbalanced data. “If you’re looking for a vessel in the ocean, you have millions of pixels of water, but very few of the vessels. These challenges are both technical and conceptual,” Prof. Svoray explained.
Addressing this requires refining AI models to better differentiate between essential objects and background noise. “The learning curve was very steep,” he admitted. “We had to develop new machine learning and mathematical skills, and interdisciplinary collaboration became crucial.”
Prof. Svoray envisions his research shaping broader environmental efforts. While AI research often focuses on improving neural networks, there remains a gap in applying these advancements to real-world problems. His lab seeks to bridge this divide by focusing on critical environmental indicators such as urban expansion, atmospheric pollution, and even glacier monitoring.
Furthermore, AI is not just transforming research—it’s reshaping education. “Students don’t code alone anymore; they work with AI tools like ChatGPT,” he noted. This evolution means that academia must adapt, redefining the role of researchers and educators in an AI-driven world. “Machines show us that they can do things, in some cases, better than us. We have to switch our direction—both in teaching and in how we approach scientific questions,” Prof. Svoray emphasized.
This shift in education highlights a fundamental change in how future scientists will be trained. Instead of focusing solely on manual data analysis, students must develop skills in AI model interpretation, algorithm refinement, and interdisciplinary problem-solving.
While some fear AI will replace human researchers, Prof. Svoray sees it as an amplifier of human cognitive ability. “AI is more than just a tool—it’s an active participant in discovery,” he explained. “It helps us process enormous amounts of data, allowing us to ask better questions and focus on higher-level problem-solving.”
However, he also warns that AI is still evolving. “We do not fully understand it yet,” he admitted. “There’s still much to learn about how AI makes decisions and how we can refine it to be more effective.” This ongoing challenge means that AI development must be accompanied by rigorous validation and ethical considerations.
As AI continues to evolve, researchers like Prof. Svoray ensure this technology is harnessed for the greater good. This unlocks new possibilities for sustainability, climate resilience, and a deeper understanding of our planet’s ever-changing ecosystems.
"We must better understand the role of humans in a world increasingly controlled by AI," he noted. "It's not just a philosophical challenge—it’s a practical one. We need to redefine how we integrate AI into education, research, and decision-making."
This article was written in collaboration with Ben-Gurion University of the Negev.