Earth is facing numerous existential environmental challenges, including crises in climate, biodiversity, and pollution. These challenges require immediate attention and innovative solutions to mitigate their impact on the planet. However, advances in environmental science are hampered by our inability to analyse ever-increasing volumes of environmental data and generate robust predictive models with actionable insights and applications. In contrast, artificial intelligence (AI) and machine learning (ML) have experienced a series of dramatic breakthroughs in recent decades. These breakthroughs have the potential to drastically enhance environmental research and enable novel approaches to tackling environmental challenges.
Yet, environmental scientists often lack expertise in data sciences, limiting their ability to leverage AI and ML tools effectively. In contrast, data scientists typically lack the domain knowledge in environmental science, hindering their ability to tackle frontier environmental challenges. This siloed training creates a bottleneck for UK leadership in science, innovation, and entrepreneurship in this emergent space. Hence, a strong interdisciplinary training need exists at the interface between the environment and AI.
The Intelligent Earth CDT will deliver a targeted, cohort-based training programme, equipping a new generation of PhD students to tackle some of the most pressing environmental issues using AI and ML through five closely connected themes:
(1) Climate
(2) Biodiversity
(3) Natural hazards
(4) Environmental solutions
(5) Core AI/ML research on complex environmental data.
The Intelligent Earth CDT will be intrinsically interdisciplinary, from tailored training for environmental science and data science streams, and interdisciplinary projects across the interface of AI and environmental sciences, all the way to interdisciplinary and intersectoral co-supervision for each student. Each student will have an external advisor from one of the non-academic partners, who will serve as host for a secondment. This setup will provide an unrivalled student experience.
The CDT will deliver a targeted interdisciplinary cohort-based training programme with two entry streams, one for highly numerate candidates from environmental, physical and mathematical science backgrounds and the other for environmentally driven candidates from computer science, data science, statistics backgrounds.
The teaching model for all courses will be tailored towards training graduate students to become independent researchers with a high degree of transferable skills. For each course, after introductory lectures, students will be introduced to the corresponding AI tools, frameworks and environmental datasets to apply the taught material in tutorial-based project work. Students will work in interdisciplinary groups tackling grand challenges in environmental science of increasing complexity with AI, supported by peer learning.
Co-creation across disciplines and sectors is at the core of the CDT with active partner involvement in each student project. This is further facilitated by a strong set of industrial and non-academic partners: IBM, NVIDIA, Google, DeepMind, ESA, Planet, Met Office, Frontier Development Lab, UK Centre for Ecology & Hydrology, National Centre for Atmospheric Science, On the Edge, Natural State, ConservationXLabs, Satellite Applications Catapult and the African Institute for Mathematical Sciences.
The in-depth training programme will train a new generation of quantitative environmental data scientists equipped to make substantial contributions in environmental and data sciences, trained in EDI and Responsible AI as well as being prepared for a wide range of career paths supported by dedicated training in enterprise and innovation.
|