Call for Proposals:
Climate Change AI Innovation Grants 2023
- Grant amount: Up to USD 150K per proposal, for projects of 12 months in duration. We will award a total of at least USD 1.2M in grants across all projects.
- Scope: Research projects at the intersection of AI/machine learning and climate change.
- Eligibility: Principal Investigator must be affiliated with an accredited university in an OECD Member country (see OECD countries). There are no eligibility restrictions on co-Investigators.
- Proposal submission deadline: March 1, 2023 at 23:59 (Anywhere on Earth time, UTC-12)
- Submission site: https://cmt3.research.microsoft.com/CCAIGrants2023
- Contact: firstname.lastname@example.org
The purpose of this grant
Artificial Intelligence (AI) and machine learning (ML) can help support climate change mitigation and adaptation, as well as climate science, across many different areas, for example energy, agriculture, forestry, climate modeling, and disaster response (for a broader overview of the space, please refer to Climate Change AI’s interactive topic summaries and materials from previous events). However, impactful research and deployment have often been held back by a lack of data and other essential infrastructure, as well as insufficient knowledge transfer between relevant fields and sectors.
The relationship between AI and climate change is also nuanced, and can manifest in various ways that either contribute to or counteract climate action. Thus, the use of AI for climate action must be performed responsibly, and ideally with quantifiable impacts.
With the support of the Quadrature Climate Foundation and DeepMind, we are excited to announce funding of USD 1.2M for projects at the intersection of AI and climate change. We are also grateful to Future Earth International for serving as the fiscal sponsor for this program.
This program will allocate grants of up to USD 150K for conducting research projects of 1 year in duration. Research projects shall leverage AI or machine learning to address problems in climate change mitigation, adaptation, or climate science, or shall consider problems related to impact assessment and governance at the intersection of climate change and machine learning.
Along with the project, the grantees must publish a documented dataset (or simulator), which was created by collating, labeling, and/or annotating existing data, and/or by collecting, simulating, or otherwise making available new data that can enable further research. We require the dataset to comply with the FAIR Data Principles (Findable, Accessible, Interoperable and Reusable).
Projects are expected to result in a deployed project, scientific publications, or other public dissemination of results, and should include a carefully considered pathway to impactful deployment. All grant IP — e.g., the dataset/simulator produced and (if applicable) trained models or detailed descriptions of architectures and training procedures — must be made publicly available under an open license.
Relevant research includes but is not limited to the following topics:
- ML to aid mitigation approaches in relevant sectors such as agriculture, buildings and cities, heavy industry and manufacturing, power and energy systems, transportation, or forestry and other land use
- ML applied to societal adaptation to climate change, including disaster prediction, management, and relief in relevant sectors
- ML for climate and Earth science, ecosystems, and natural systems as relevant to mitigation and adaptation
- ML for R&D of low-carbon technologies such as electrofuels and carbon capture & sequestration
- ML approaches in behavioral and social science related to climate change, including those anchored in climate finance and economics, climate justice, and climate policy
- Projects addressing AI governance in the context of climate change, or that aim to assess the greenhouse gas emissions impacts of AI or AI-driven applications, may also be eligible for funding. (Studies addressing this area may be exempt from the dataset publication requirement.)
For context, a list of the research projects funded during the 2022 Innovation Grants cycle is available here.
Proposals focused on using AI/ML to address climate change mitigation in the electric power sector (including, but not limited to Optimal Power Flow and related multi-level problems like Unit Commitment) may also optionally request support from a DeepMind Engineer, in addition to the financial award. Applicants who may be interested in taking advantage of this option will be asked to indicate this in the CMT submission form.
Each application must have a Principal Investigator (PI) who is affiliated with an accredited university in an OECD Member country. The PI must be eligible to hold grants under their name at their accredited university; this may include, e.g., faculty, postdocs, or research scientists (depending on the institution). There are no eligibility restrictions on co-Investigators, and multi-country and multi-sectoral collaborations are encouraged (e.g., including members outside OECD Member countries or from non-research institutions).
Current members of the Climate Change AI Board of Directors and Climate Change AI staff cannot apply to this grant as a PI, and they may not receive funds towards their own salary. Program Chairs and Meta-Reviewers for this grant may not apply or receive funds in any way (however, Reviewers may, and conflicts of interest will be appropriately managed during the review process).
We do not fund research activity that is currently funded by other grant programs. If other grant proposals for the same project have been submitted and/or are under consideration, the relation of the present proposal to those other proposals needs to be clearly explained. If the proposal is selected for funding, no aspect of a project should be double funded by other funding bodies.
|Call release date||Nov 2, 2022|
|Proposal submission deadline||Mar 1, 2023|
|Notification of results||May 15, 2023|
|Award start date||Aug 15, 2023|
|Award end date||Aug 15, 2024|
Proposals will be reviewed through a single-blind process by independent reviewers.
Projects will be evaluated on the following criteria:
- Climate relevance: Projects should demonstrate a clear link to climate change mitigation and/or adaptation. Given the cross-cutting nature of climate change, this can include a wide range of topics with which climate change interacts and intersects, but the relationship to climate change should be made explicit.
- AI/ML relevance: Projects should employ or address AI or ML in a way that is well-motivated and well-scoped for the problem setting. This includes both projects where AI or ML are a central component, as well as those where AI or ML are one among many components. Projects proposing the implementation of AI/ML techniques will not be penalized if other techniques or approaches are found to be better-suited as the project progresses; negative results are welcome if well-tested.
- Dataset: The proposed dataset or simulator to be created should serve to enable further impactful work at the intersection of climate change and machine learning beyond the project being proposed. We require the dataset to comply with the FAIR Data Principles (Findable, Accessible, Interoperable and Reusable).
- Pathway to impact: Proposals should address how their work, if successful, can be deployed or implemented in practice to aid climate mitigation and/or adaptation. This can be addressed in the form of deployments planned as part of the project itself, or via a concrete plan for disseminating the work among relevant sectors or organizations.
- Ethics: Proposals should explicitly discuss ethical considerations and implications of their work. This includes discussion of relevant stakeholders and equity considerations of the problem addressed, as well as the scope and potential negative social or environmental impacts of the proposed solution, including how these risks will be avoided or mitigated in the project’s execution. (See, e.g., the NeurIPS ethics guidelines for a discussion of ethical considerations pertinent to ML.)
- Feasibility: The scope of the proposed project should be realistic with respect to the associated timeline and budget.
- Expertise of team: The proposed team should have demonstrated expertise in areas of relevance to the development and execution of their project, notably the relevant area(s) of climate change mitigation and adaptation and in AI/ML. Interdisciplinarity and diversity within the proposed team will be viewed favorably.
In addition, the following aspects will be considered favorably during the review process:
- Deployment partners: Project teams including relevant organizations through whom the proposed work could be impactfully deployed will be viewed favorably.
- Traditionally under-funded areas of work: Projects that are impactful but may not be traditionally covered through other funding streams will be given priority as part of this call. Examples include projects that may not fit neatly into one discipline or area of study, or projects serving stakeholders with limited access to capital.
- Equity: Projects that explicitly incorporate equity-related considerations — e.g., through the choice of problem addressed, or stakeholders that are partnered with — will be viewed favorably.
Across the full cohort of grantees, we will additionally seek to allocate grants to represent multiple sectors of climate change mitigation and adaptation, as well as coverage across many geographic regions.
All applications must be received by March 1st, 2023 at 23:59 (Anywhere on Earth time, UTC-12). Applications should be made via the CMT website, which will require the following information.
Basic information. The CMT submission portal will require the title and abstract of the proposal; the name, affiliation, and country of affiliation of the Principal Investigator; the names, affiliations, and countries of affiliation of any co-Investigators; and additional short declarations about the project. The first name in the CMT author list will be treated as the Principal Investigator. Only one Principal Investigator may be named, but there is no limit on the number of co-Investigators. Please note that the institution of the Principal Investigator will be used to determine eligibility, and will be responsible for receipt and any further distribution of the funds if a grant is awarded.
Project Description. A detailed description of the project (maximum 12 pages including figures/tables, using no smaller than 12pt font size, single line spacing, and 1 inch margins), with unlimited additional pages allowed for references. The Project Description should be submitted as one PDF attachment via CMT, and include the following subsections (please use the same order and headers to separate the subsections):
- Project title, the name and affiliation of the Principal Investigator, and the names and affiliations of any co-Investigators.
- Summary: A short description of the proposed project of up to 250 words.
- Research Outline: A detailed description of the proposed project. This section should address both the proposed methodology (e.g., machine learning) and application area (a climate change-relevant topic), and should explicitly address what gap the proposed project fills in climate change mitigation or adaptation, as well as why the proposed methodology is useful and appropriate for addressing this gap.
- Deliverables: A description of what concrete deliverables (e.g. papers, code, datasets, deployed systems) are expected from the project.
- Timeline: A timeline for key milestones of the project, aligned with the deliverables described above.
- Team: A description of the relevant expertise of each team member and how it relates to the project.
- Pathway to Impact: A plan for how the proposed work will have an impact on GHG emissions or societal resilience to climate change. This should be as specific as possible. It is not required that deployment take place within the duration of the project, but all projects should be scoped and developed in such a way as to facilitate impactful deployment in future. At a minimum, this section should address: how the authors plan to engage with end users/other relevant stakeholders during the project, which stakeholders will make use of this work, how exactly it will be useful for these stakeholders, and considerations that are necessary to facilitate impactful deployment (bearing in mind the potentially different incentives for various stakeholders involved).
- Dataset Plan: All projects must propose a new dataset that will be created and made publicly available in compliance with the FAIR Data Principles (Findable, Accessible, Interoperable and Reusable). “Creation” of a dataset may include annotating data with labels, collecting completely new data, collating existing data from multiple sources, creating a data simulator (e.g. for reinforcement learning) that is well-grounded in reality, or open-sourcing existing data that was formerly private. This section of the Project Description should describe the dataset, what it will contribute (as compared to existing datasets), and what will be done to create the dataset. The description should also include a detailed plan for how the data will be documented, shared and preserved, in particular elaborating in detail how compliance with each of the FAIR Data Principles (Findable, Accessible, Interoperable and Reusable) will be ensured. Note that teams will be required to use datasheets to document their created datasets.
- Equity Considerations: This section should describe equity-related considerations related to the project, and how the team will shape the project with these in mind. This discussion may include the nature of the research, composition of the team, and/or nature of the stakeholders outside the team who will be worked with.
- Ethical Considerations: A description of any broader ethical considerations associated with the development and deployment of the work, including but not limited to those connected to climate change. This section should include a description of potential societal impacts or side effects, as well as factors to bear in mind to mitigate negative effects, including important stakeholders to include.
Budget and Budget Justification. An itemized Budget (1 page) indicating the total amount requested and how these funds will be used if a grant is awarded, and a brief Budget Justification (1 page) of these amounts, submitted as one PDF file through CMT. Eligible expenses include salaries for Investigators, students, and other research staff; materials, equipment, software, and compute; and expenses associated with conferences and other project-related travel. The Budget should also indicate any institutional overhead, at a maximum rate of 10% of the total amount requested. If this project has other sources of funding, the Budget should make clear which research activities are proposed to be funded by the present grant, and which research activities are funded by other sources. Please note that funds will be contracted solely to the accredited university with which the Principal Investigator is affiliated; any further dissemination of funds to partner institutions must be managed by the lead institution.
We encourage you to use this Budget Template and adapt it to your project needs by adding or subtracting lines and/or columns to it.
CVs of key personnel. CVs for the Principal Investigator and all co-Investigators, as a single PDF file (no page limit).
About Climate Change AI
Climate Change AI is a nonprofit initiative to catalyze impactful work at the intersection of climate change and machine learning. Since it was founded in June 2019, CCAI has inspired, informed, and connected thousands of individuals from across academia, industry, and the public sectors, through its foundational reports on AI and climate change, networking and knowledge-sharing events, educational initiatives, and global grants programs. See our website for further details.
Q: What counts as an OECD Member country under the eligibility criteria of this grant?
A: This refers to the 38 OECD Member countries listed on the OECD website. These countries are: Australia, Austria, Belgium, Canada, Chile, Colombia, Costa Rica, Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Iceland, Ireland, Israel, Italy, Japan, Korea, Latvia, Lithuania, Luxembourg, Mexico, Netherlands, New Zealand, Norway, Poland, Portugal, Slovak Republic, Slovenia, Spain, Sweden, Switzerland, Türkiye, United Kingdom, United States.
Q: I am not an AI or ML expert; can I apply?
A: Yes, as long as your project includes an aspect of AI/ML which addresses one of the areas described in “Purpose of this grant”. You may want to consider finding an AI or ML expert to collaborate with; our workshops, online discussion platform, happy hours, and community directory could be helpful for this.
Q: I am not a climate expert; can I apply?
A: Yes, as long as your project addresses a problem of climate change, with a pathway to impact clearly described. You may want to consider finding an expert in the relevant climate-related domain to collaborate with; our workshops, online discussion platform, happy hours, and community directory could be helpful for this.
Q: I’m from a non-OECD Member country but currently at an institution in an OECD Member country; can I apply?
A: Yes, as the funds will be disbursed through your institution, not to you directly.
Q: I’m from an OECD Member country but currently at an institution in a non-OECD Member country; can I apply?
A: At this time, you unfortunately cannot apply as a Principal Investigator; for logistical reasons, we are currently only able to disburse funds to institutions in OECD Member countries. However, you may participate as a co-Investigator in a grant proposal, provided the Principal Investigator meets the eligibility requirements. Our workshops, online discussion platform, happy hours, and community directory could be helpful in finding collaborators.
Q: I’ve been affiliated with Climate Change AI in the past, been a co-author with members of Climate Change AI, or otherwise involved with Climate Change AI. Can I apply?
A: Yes, except under specific circumstances. Specifically, Program Chairs and Meta-Reviewers for this grant may not apply or receive funds in any way (however, Reviewers may). Current members of the Climate Change AI Board of Directors cannot apply to this grant as a PI, and they may not receive funds towards their own salary. Other Climate Change AI affiliates are welcome to apply in any capacity.
Q: Am I eligible to apply for these funds if I have applied to other sources for the same research activity?
A: The exact same research activity cannot be double funded. However, this grant may be used to fund a component of a project whose other components are under consideration or have received funding from other sources. This structure should be clearly described in your budget.
Q: Can I apply multiple times with different projects?
A: Yes, you are welcome to apply multiple times. However, as mentioned above, we will seek to select a cohort of grantees “to represent multiple sectors of climate change mitigation and adaptation, as well as coverage across many geographic regions.” This may in turn reduce the probability of multiple proposals from the same team being funded.
Q: I applied for the 2022 Innovation Grants program and my proposal was rejected. Can I apply again with the same proposal?
A: Yes, you are welcome to apply with the same proposal. However, make sure to include relevant updates, including an updated budget and justification.
Application and review process
Q: Can I send my application via email?
A: No, all applications must be via CMT.
Q: Do I need to use a particular software, like LaTeX or Word, to write the proposal?
A: No, you may use any software to write the proposal, as long as it follows the requirements on length, font, line spacing, and margins.
Q: Is review of the proposals double-blind?
A: No, the review process is single-blind (reviewers’ identities are hidden from proposal authors). Proposals are encouraged to be very specific about their pathway to impact, and this is likely to contain de-anonymizing information that reviewers would need in order to evaluate the feasibility of the proposed project.
Q: I’m uncertain about the start and end dates for my project, what should I do?
A: Just give your best guess, with an explanation of the reasons for your uncertainty if you believe it would help in evaluating your proposal.
Q: By when do I need to publish my dataset?
A: By the end of the year-long grant, there should be a well-defined plan for data release, with data released no later than one year after the completion of the grant.
Q: My project would require a budget greater than the maximum allowed (USD 150K). What should I do?
A: In order to distribute grants equitably and fund a larger number of projects, we will not allocate more than USD 150K to a single project. You should describe and apply to us for a USD 150K portion of your project, and apply for additional funding elsewhere. Make sure to describe this in your budget, including the additional funding sources you have secured or intend to apply to.
Q: My institution takes an overhead greater than 10% of the grant. Am I still eligible to apply?
A: You are still eligible to apply, but you will need to obtain an exemption from your institution regarding overhead, as your institution will not be allowed to take more than 10% overhead.
Q: How is the 10% cap on overhead defined?
A: The overhead should be at most 10% of the total amount requested, and this overhead amount should be internal to the total budget requested. For example, if the total budget proposed is $150K, then at least $135K must be direct project costs, and at most $15K can be overhead.
Q: May I use my own format for the budget, rather than using the provided template?
A: Yes, you may.
Scope and relevance
Q: Does this grant call include other environmental or social issues that do not directly pertain to climate change?
A: All proposals should clearly describe the relevance to climate change, as well as the pathway to impact for the climate problem. As problems of climate change intersect with a host of other issues, we welcome grantees to lean into these connections and consider their project holistically.
Q: Does the machine learning proposed in the project need to be ‘novel’?
A: No, in the sense that it is perfectly acceptable to use a previously published ML algorithm or architecture. However, the scientific knowledge generated in this project (e.g. trying the previously published ML technique in a novel setting, combining existing techniques in a novel way, etc.) should be novel, i.e., informative and not previously published.
Q: The pathway to impact for my project is highly speculative. Will this hurt my proposal?
A: We encourage submissions anywhere on the spectrum from guaranteed-but-small impact to high-risk/high-reward. The important part for evaluation is that you thoroughly and accurately describe the pathway to impact, including your level of uncertainty about any aspects, and take steps to reduce or address uncertainty where possible. E.g., it may hurt your proposal if the speculation is due to lack of domain research, but not if it is due to irreducible uncertainty about future outcomes, physical processes, etc.
Q: Does the dataset have to be for supervised learning?
A: Not at all. In order to enable future impactful work, do ensure that you clearly describe the way you intend the dataset or simulator to be used.
Q: My proposed project involves human or animal experimentation that requires explicit ethics approval. Does getting the grant provide this approval?
A: No. You should include in your project description (and budget, if appropriate) any approvals or regulatory oversight necessary for your project, and obtaining those are your responsibility.
Q: What is climate change mitigation?
A: Climate change mitigation refers to the reduction of greenhouse gases in order to reduce the extent of climate change. As described by the IPCC Working Group III, this “is achieved by limiting or preventing greenhouse gas emissions and by enhancing activities that remove these gases from the atmosphere.” For examples of where AI and ML can help with climate change mitigation, please see Climate Change AI’s report on “Tackling Climate Change with Machine Learning.”
Q: What is climate change adaptation?
A: Climate change adaptation refers to activities that aim to prepare for or build resilience to the conditions created by climate change. For more information, please see resources from the IPCC Working Group II. For examples of where AI and ML can help with climate change adaptation, please see Climate Change AI’s report on “Tackling Climate Change with Machine Learning.”
Q: What is climate science?
A: Climate science is the study of the environmental processes that determine past, present, and future climate. For more information, please see resources from the IPCC Working Group I. For examples of where AI and ML can help with climate science, please see Climate Change AI’s report on “Tackling Climate Change with Machine Learning” as well as the proceedings of the Conference on Climate Informatics.
Q: What is meant by AI and ML?
A: Artificial intelligence (AI) refers to any algorithm that allows a computer to perform a complex task — typically, tasks such as speech, perception, and reasoning that are associated with human intelligence. Machine learning (ML) is a sub-area of AI referring to techniques whose behaviors or outcomes depend on “learning” — corrections or changes made as a result of seeing examples or descriptions — rather than being hard-coded in advance. ML is used to describe a wide variety of techniques that range in their complexity, including, e.g., linear regression, decision trees, and deep neural networks.
Q: The project evaluation criteria refer to “equity.” What is meant by “equity” in this context?
A: The word “equity” in this case refers to considerations of diversity, equity, and inclusion (rather than, e.g., the financial meaning of the term).
Q: I have a question that isn’t answered here. What should I do?
A: Please contact us at email@example.com.