Accelerating climate science with AI

Go back to programme

Accelerating climate science with AI

  • Watch

    * Register (or log in) to the AI4G Neural Network to add this session to your agenda or watch the replay

    • Watch

      * Register (or log in) to the AI4G Neural Network to add this session to your agenda or watch the replay

    Climate change is one of the most pressing global challenges, with human emissions now unanimously accepted as the key driver. However, significant scientific uncertainties remain regarding the Earth’s climate sensitivity, which directly translate into uncertainties in the remaining carbon budget on the path to achieving the targets of the Paris Agreement. Current advances in climate science are hindered by the limitations of the methodologies commonly used for scientific data analysis, some of them ill-equipped for capturing complex relationships present in the huge volumes of data now available. At the same time, the field of machine learning has seen a series of dramatic breakthroughs over the past decade, driven by the development of new data-intensive algorithms.

    A kick-off event on 7 July 2021 will set the stage for a series of expert talks over the following months which will bring together leading voices from academia, industry and policy to provide a forum to ensure climate science is able to capitalise on the latest advances in AI and machine learning for the benefit of humanity. The outcome will be a whitepaper proposing ways forward.

    Keynote Remarks at the ITU AI and Climate Change kick-off event, by Ban Ki-Moon

    Introductory remarks on “Accelerating climate science with AI”, by Philip Stier

    AI expands scope of climate science; ITU News blog, 13 July 2021

     

    Shownotes: 

     

    [01:41] Introduction by Fred Werner (ITU) 

     

    [05:21] Introduction by Philip Stier

    -Host of the Climate Science series

     

    [05:28] Climate Science

    -Climate Change is one of the most pressing global challenges. 

    -Human emissions now unanimously pressing global challenges

    Significant scientific uncertainties remain, in particular regarding the impact on extreme climate events. 

     

    [05:47] AI and ML

    -AI and machine learning have seen a series of dramatic breakthroughs. 

    -Applications are transforming scientific areas. 

    New algorithms

     

    [06:21] Accelerating Climate Science with AI 

    -Identification and quantification of complex association between features in a vast model and earth observation datasets. 

    -Climate models are skillful . 

    -Small dataset of earth has been used. 

     

    [07:16] Programme 

    Mr. Ban Ki Moon

    Ms. Valerie Masson Delmotte 

    Mr. Yoshua Bengio 

     


     

    [08:15] Introduction by Mr. Houlin Zhao 

    -Applications of AI to accelerate the SDG’s. 

    -How can we accelerate climate science with AI? 

    -Introduction to Mr. Ban Ki Moon

    -Global impact demands new partnerships and AI for good will works as the platform for this. 

    -Today's session means an introduction to the new topic on how accelerate climate change 

    -Introduction to Mr. Bengio

     


     

    [16:16] Introduction by  Mr.Ban Ki-Moon 

    -How can problems be tackled with the use of AI? 

    -International community is combating climate change and is looking for effective ways to do it. 

    -Covid 19 has changed our lives, but climate change is still worsening. 

    -Highlight the use of technology (AI) to have a brighter future for our humanity.

    -How can we mitigate the threats posed by Climate Change

    -Global efforts are important through multilateralism and partnerships.

    -I have prioritised, as secretary-general, climate change

    -Innovation and technology should be in the center of the fight against climate change (global goals: food security, energy, etc). 

    -Mindset is important to combat climate change. 

    -Science, innovation and technology is important to achieve SDGs and it is something that the scientific community should have in mind. 

     


     

    [35:45] Mr. Houlin Zhao summary of Mr. Ban Ki-Moon speech

    -Support on the use of ICT’s from Mr. Ban Ki-Moon.

    -The United Nations system has a framework on using ICTs for the development through the World Summit on the Information Society (WSIS)

    -ICT’s have helped societies during COVID-19 situation.

     

    [38:12] Mr. Ban Ki-Moon 

    - When it comes to climate, there is no time to lose. 

    -Covid-19 is not an independant situation of climate change. 

     

    [42:17] Mr. Philip Stier question 

    -Advice on how to translate the crisis into specific actions? 

    We need political leadership. With this, a partnership with civil society the world will have a better response against these crises.

     


     

    [46:16] Introduction by Valerie Masson-Delmotte

    -IPCC reports climate change affects many aspects of life, such as food, security, etc. 

    -New approaches to identify new climate projects and plan a future answer. 

    -Clouds play a key role in the response of the climate system. 

    -Data available is important to use AI to tackle climate problems: quality and quantity. 

     

    [52:48] Land

    -Land management is important to have food security. 

    -Climate and biodiversity are very linked. Loss degradation of the ecosystem is made by carbon emissions. 

     

    [54:35] Ocean and Cryosphere

    -There are no linear responses associated with these two fields. 

    -Bring together solutions aligned to these is needed. 

    -How to do the evaluation of the local and regional information.

    -Now, climate change events are being presented more repeatedly. 

    -Coordinate initiatives are needed and it's important to understand the nature of every climate event. 

    -Science can inform action to address climate change problems. 

    -Innovative approaches are needed at the international level, to harmonize an action.

     

    [01:00:31] IPCC report

    -it’s an online tool to look at past, present and future events. 

    -75 reviews from the scientific sector, which is exhausting.

    -AI techniques can facilitate review of this report.

     

    [1:03:35] Introduction from Philip Stier to Mr. Bengio

     

    [1:04:40] Introduction by Mr. Yoshua Bengio

     

    [1:04:50] AI and fighting climate change

     

    [1:05:21] Environmental Application of AI 

    -Optimizing energy resources.

    -Climate modelling: predicting climate change effects. 

    Accelerate R&D of new material. 

    -Visceralization: visualize the future. 

     

    [1:06:23] Climate change + AI (CCAI) 

    -Tackling climate change with machine learning paper. 

    -Connect xperts outside of AI. 

    -A grid connects the application of AI in different areas. 

     

    [1:07:47] Global Partnership on AI Climate and Biodiversity Committee

    -A responsible AI strategy for the environment

    -Help identify areas for investment in AI that can contribute to the Paris Accord and to the preservation of biodiversity. 

     

    [1:08:43] Synthesizing new material for batteries in carbon

    -Areas of application: Synthesizing new materials.

    -The process comprises two aspects: Modelling aspect and search aspect.

    -Both aspects can be done by using AI and will help to solve problems in their development. 

     

    [1:09:26] Learning to explore the space of material 

    -Exploratory agent -> Imagined experiment -> World model -> learning declarative knowledge -> updated world model

    -Mentioned process will be shorter with AI

     

    [1:10:36] Physics-Constrained ML Models for Climate or Material 

    -Combination of Climate Science, Physics and other fields with ML

    -ML approximate physical model dynamics faster to enable greater space-time resolutions. 

    -Hybrid models: incorporate physical invariantes and equivariance in dynamic models. 

    -Use deep learning to discover new latent abstract space where dynamics can be run at low resolution 

     

    [1:12:02] Why insufficient action? psychological factors? 

    -Human cognitive biases: we underestimate the threat of climate change.

    -Far away in time and space and abstract: we cannot see the future effects of climate change. 

    -It is hard to compete with short-term economic issues.

    -Information bubbles/biases (information filtered). 

     

    [1:13:21] This Climate Does not Exist

    -Extreme climate events are occurring all around the planet and climate change highlights the contribution to them.

     

    [1:14:01] The Climate Change Visualization Project

    -Project: generate images simulating climate change effects. 

     

    [1:15:01] Bring Future Societal Impact into Prices Today 

    -Carbon pricing should work, but governments are not setting the price high enough because of cognitive biases (remote threats vs. immediate loss). 

    -Defer to the later government to decide what the price should have been. 

    -Align carbon price with our current best rational bet of what future government will charge, and include the cost of uncertainty. 

     

    [1:16:59] Bring the Future Prices Today with Retroactive Carbon Pricing

    -Example: Image new law 2022 on retroactively applying carbon price to emission from 2022 onwards. 

    -2023 If price increases by 10, polluters of 2022 owe 10 extra. 

     

    [1:19:52] Bring future societal impact into prices today 

    -To avoid uncertain cost, companies buy an insurance contract

    -The insurance contracts are market-traded: intencetive for honest forecasts. 

    -Fund used to compensate the underprivileged. 

     

    [1:22:03] Closing 

     

    [1:22:31] Which areas do you think we can contribute the most? There are particular directions. Spatial models, images, and others.

     

    [1:23:12] What kind of AI exists for the plastic problem in the ocean? I don’t know about this in particular. Collecting data is the most difficult thing, so managing data should be addressed to solve this problem 

     

    [1:24:56] Girls and women in AI. How to solve the problem of gender equality in the field of AI? Mentorship at early ages. Emulation and motivation of kids.  

     

    [1:27:07] Mindset. Education is important. Travel in time. Go to the past and see the present and the effects of the past to look forward to the future. This is a powerful tool. 

     

    [1:29:23] AI for Good next session Duncan

    -16 July we will take a deeper dive into the many ways AI can accelerate climate science. This will lead towards the development of a whitepaper. 

     

    [1:30:21] Closing remarks by ITU. 

     

     

    Share this session
    In partnership with