Our Grants, Investments, and Regrants

This regrant will support a summer research internship at Columbia University in Barnard’s Programming Languages Lab under Prof. Santolucito, as well as equipment and conference attendance. The research project is aimed at modeling and aligning program synthesis models in the setting of computational creativity.

We recommended a grant over three years to support OWID’s work producing high quality data and analysis of global trends like the rise in living standards and effects of COVID-19. This grant will specifically support work tracking trends that are relevant to humanity’s long term prospects.

We recommended a grant to allow GPI to hire pre-docs and post-docs from economics, philosophy, and psychology and publish analysis on how to do the most good.

We recommended a grant to run two pilot studies in order to develop a platform that will allow policymakers to rapidly get reliable estimates about important quantitative parameters to inform policy measures during times of crisis, like during a pandemic. The pilot will focus on evaluating the cost-benefit profile of non-pharmaceutical interventions during pandemics.

We recommended funding to support their longtermist grantmaking.

We recommended a grant to an organization aiming to reduce childhood lead exposure and eliminate childhood lead poisoning worldwide.

We recommended a grant to support the creation of a small research organization that will use probabilistic models to forecast important questions about the future, in particular involving transformative artificial intelligence.

We recommended a grant to create a nuclear winter food scale-up plan which governments can adopt.

We recommended a grant to support Longview Philanthropy creating a longtermist coworking office in London.

We recommended a grant to support 18 months of operations for a longtermist coworking space in Berkeley

We recommended a grant to support researcher salaries, funding for contractors for evaluating AI capabilities, projects like the ELK prizes and workshops, and administrative expenses.

This regrant will support six months of salary for Charbel-Raphaël Segerie to work on AI safety mentoring and study.

This regrant will fund the writing of a book about classical utilitarianism and its foundations and implications.

This regrant will fund the prizes for a competition on using machine learning models to estimate when human values are in conflict and estimate moral uncertainty.

This regrant will support Michael Trazzi to make 20 episodes of his podcast and YouTube channel about research on AI and its risks called "The Inside View." Funds compensated Michael for his time, for equipment, and for marketing.

This regrant will support three months of independent AI interpretability research.

This regrant will support Professor Cihang Xie's research on adversarial robustness and honest AI.

This regrant will support funding for students at Harvard and Cornell to establish student groups to learn about and research longtermist space governance, including events, stipends, and prizes.

This regrant will support three years of work on the Global Priorities Research psychology research agenda, for a post doc, conference attendance, research assistants, and payments for experimental participants, among other expenses.

This regrant supports work investigating FarUVC, a wavelength of light which could safely sterilize pathogens in occupied rooms. This regrant supports the International Commission on Non-Ionizing Radiation Protection, a nonprofit group that sets international exposure guidelines, to 1) determine what additional information is needed to re-evaluate the guidelines for 180-230nm light, and 2) specify experiments that would provide that knowledge, including methodological details.

This regrant will cover the costs of running a seven month AI safety research fellowship for philosophy academics. The funds will cover the costs of travel, housing stipend, office space, guest speaker honorariums, and the hiring of researchers, advisors, a project manager, and organizers for the seven month period.

This regrant will support the design and creation of a tabletop roleplaying game in which simulates potential risks from advanced AI.

This regrant will support an independent researcher to investigate the magnitude of AI risk and the most useful actions to take to address it, how to improve Effective Altruist decisionmaking according to "metarational" principles, and potentially valuable projects in metascience.

This regrant will support six months of independent research on interpretability and other AI safety topics.

This regrant will support the hiring of a PhD student or postdoc to help orient their lab toward AI safety.

This regrant will fund their research agenda into robust generative models in the face of non-randomly missing data, targeting use cases in healthcare, materials science, and climate.

This regrant will support leadership coaching for leaders of Effective Altruist projects.

This regrant will support ten fellows to work on mentored AI Governance projects over the summer.

This regrant will support a teaching buyout which will allow Jenny to spend 20 hours a week on AI Governance research.

This regrant will support several research directions in interpretability for 2-3 years, including: empirical evaluation of knowledge and guessing mechanisms in large language models, clarifying large language models’ ability to be aware of and control use of internal knowledge, a theory for defining and enumerating knowledge in large language models, and building systems that enable human users to tailor a model’s composition of its internal knowledge.

This regrant will support a teaching buyout to allow Yan to focus on forecasting AGI scenarios, as well as research on beneficial alignment and deployment.

This regrant will support independent study and research on AI safety and alignment alongside research assistant work for Lennart Heim.

This regrant will fund Professor Tim Oates and a graduate student to work on AI alignment research.

This regrant will support a year of coordinating several EA mentoring and community building activities with South American Spanish speakers.

This regrant will support four months of theoretical AI safety research under Evan Hubinger’s mentorship.

This regrant will support six months of AI safety work.

This regrant will fund Professor Pearce’s graduate students to conduct research projects on resilient food, specifically on manufacturing rope for seaweed farming and on making leaf protein concentrate.

This regrant will fund prizes for papers at a NeurIPS Workshop, with $50k for papers that have the best discussion of AI x-risk, $50k for best overall papers.

This regrant will support operations and logistics for the Harvard-MIT Existential Risk Summit held in Berkeley at the end of July. This four-day retreat will bring together incoming Harvard and MIT first years with Berkeley EAs and members of the Harvard/MIT EA communities.

This regrant will support a small grants program managed and advertised via Hear This Idea which will give funding to people to launch podcasts or audio projects focused on ideas that matter.

This regrant will support the development of a first draft of a resilience plan for nuclear winter/extreme pandemic.

This regrant will support a year of work translating popular content on Effective Altruism into Spanish.

This regrant will support general expenses for running the Future Forum conference.

This regrant will support six months of research on technology policy case studies of key historical figures who influenced technology development.

This regrant will support a career transition grant from physics to AI safety.

This regrant will support historical research contracting work.

This regrant will fund Professor Pierson to expand her program of research on fair and robustly aligned ML systems, with a focus on methods that will remain relevant as capabilities improve.

This regrant will provide a year of salary, funding to hire a research assistant, and compute for an independent AI safety researcher.

This regrant invested in two mental health, productivity, and personal growth applications.

This regrant will provide funding for six people to spend three months producing the Center’s research agenda and creating and registering the organization.

This regrant will support a career transition grant to allow David to pivot to working as a frontend developer for Effective Altruist organizations, becoming a 'language model psychologist,' or participating in charity entrepreneurship.

This regrant will allow Sam to manage a group of students to start a 'London Existential Risk Initiative' (LERI). This provided funding for that student group, and in particular will be supporting three UCL student organizers.

This regrant will support 12 months of High Impact Medicine so that they continue (and scale up) their activities.

This regrant will fund BvS, a newly set up humanist student scholarship foundation, to take on 10 more fellows.

This regrant will support four months of their work on in-depth analyses of charities.

This regrant will support Joseph Bloom to spend 3-6 months transitioning his career to EA/Longtermist work, including conducting research on biosecurity.

This regrant will allow the Montaigne Institute to hire a full time staff member to work on AI policy.

This regrant will support one year of funding to found and run a broad-interest, high-end website, featuring high-quality writing and appealing visualizations focused on promoting longer-term thinking.

This regrant will support 6 months of salary for work on biosecurity projects, partly mentored by Tessa Alexanian, Officer of Safety and Security at iGEM.

This regrant will support six months of work on distilling AI alignment research.

This regrant will fund students and postdocs to attend a conference on Human Centered AI at UC Berkeley and work on a resulting report.

This regrant will fund one year of operations for the Social Science Prediction Platform, a platform which allows economists and other social science academics to use forecasting in their research.

This regrant will support a course to teach civil servants in the UK to read, process, and use scientific research in their policy work. The course will focus on examples from emergency scenarios including pandemics and other existential risks.

This regrant invested in Aligned AI, an AI safety organization seeking bridging funding.

This regrant will support a risk-benefit analysis of virology research.

This regrant will fund the operations of the Fund for Alignment Research, which primarily provides engineering support to AI alignment researchers.

We recommended a grant to support Redwood Research, an AI safety organization, and MLAB, a Redwood project to teach machine learning to individuals interested in AI safety, for 18 months which will support salaries, primarily for their researchers and technical interns, and the rest for compute.

This regrant will support three years of work, paid out conditional on progress reports for a cybersecurity project for AI research.

This regrant will support the second “Finding the Right Abstractions” workshop, likely to take place around January 2023, which connects applied category theory and AI alignment communities in an effort to understand self-selecting systems.

This regrant will support the recording of audio narrations of important philosophical papers. The recordings will be packaged up as a podcast feed with its own website, brand and YouTube channel.

This regrant will support Simeon Campos to work for a fraction of his year on AI safety projects.

This regrant will support two researchers, Matthew Prewitt and Divya Siddarth, to create a research agenda on “collective intelligence” specifically aimed at mitigating Global Catastrophic and Existential risks, and run some trial projects based on it.

This regrant will support ongoing research and convening with representatives from UK government, academia/think-tanks, and AI industry throughout 2022-23. These activities will be supplemented by written outputs and the program will focus on UK AI policy and risk reduction, pandemic prevention policy, defense policy, cybersecurity policy, and UK navigation of great power technology relations.

This regrant will support continued mentorship of a medical researcher and work generating novel medical research literature reviews, on a variety of topics, including long covid.

This regrant will support five months of building and supporting the longtermist and AI alignment communities.

This regrant will support additional funding for Condor Camp, the program for top Brazilian university students to come to Peru for a retreat to learn about Effective Altruism.

This regrant will support a promising student who is working to bridge algorithmic information theory with epistemology and AGI research for his PhD at the University of Cambridge.

This regrant will support a 6-month trial of setting up the Global Priorities Encyclopedia (GPE), a reference work on global priorities research featuring articles by academics and researchers from the effective altruism community. Funding would pay the project lead, web developers, and article contributions.

This regrant will fund Gryphon Scientific to analyze options for and hold a workshop on protecting potential pandemic pathogen genomes.

This regrant supports the creation and initial hiring for an organization that is centered around outreach to machine learning professionals about AI safety and alignment.

This regrant is an investment in VIVID, a customization-based mindset improvement app that aims to improve peoples' self-reflection and productivity.

This regrant will support an experiment where a recruiting agency will headhunt ML Engineers for AI safety organizations.

This regrant will support a global media outreach project to create high quality video and social media content about rationalism, longtermism and Effective Altruism.

This regrant will support six months of research on topics including how legal services can be effectively provided to the Effective Altruism community, materials to be included in a legal services handbook for EA organizations, novel legal questions particular to the EA community that might benefit from further research initiatives, and ways to create an effective EA professional network for practicing lawyers.

​​This regrant will support Manifold Markets in building a play-money prediction market platform. The platform is also experimenting with impact certificates and charity prediction markets.

This regrant will support prizes for a trojan detection competition at NeurIPS, which involves identifying whether a deep neural network will suddenly change behavior if certain unknown conditions are met.

This regrant will support the creation of the Active Longtermist Emergency Response Team, an organization to rapidly manage emerging global events like Covid-19.

This regrant will support prize money for a writing contest for critically engaging with theory or work in Effective Altruism. The goal of the contest is to produce thoughtful, action oriented critiques.

This regrant will support a researcher and research assistant to work on high-skill immigration and AI policy at FAS for three years.

This regrant will support Ought’s work building Elicit, a language-model based research assistant. This work contributes to research on reducing alignment risk through scaling human supervision via process-based systems.

This regrant will fund a summer program for up to 100 students to spend 9 weeks studying machine learning, deep learning, and technical topics in safety.

This regrant will support a project to create and house operations-related resources and guidance for EA-aligned organizations.

This regrant will support an essay contest on “Accounting for Existential Risks in US Cost-Benefit Analysis,” with the aim of contributing to the revision of OMB Circular-A4, a document which guides US government cost-benefit analysis. The Legal Priorities Project is administering the contest.

This regrant will support a NeurIPS competition applying human feedback in a non-language-model setting, specifically pretrained models in Minecraft. The grant will be administered by the Berkeley Existential Risk Initiative.

This regrant will support QURI to develop a programming language called "Squiggle" as a tool for probabilistic estimation. The hope is this will be a useful tool for forecasting and fermi estimates.

This regrant will support rerunning the highly-cited survey “When Will AI Exceed Human Performance? Evidence from AI Experts” from 2016, analysis, and publication of results.

This regrant will support the creation of an AI Safety organization which will create a platform to share AI safety research ideas and educational materials, connect people working on AI safety, and bring new people into the field.

This regrant will fund the writing of Slaoui's memoir, especially including his experience directing Operation Warp Speed.

This regrant will support a project to use a combination of human feedback and Machine Learning to label misinformation and reasoning errors in popular news articles.

This regrant will support the creation of The Good Ancestors Project, an Australian-based organization to host research and community building on topics relevant to making the long term future go well.

This regrant will support three prizes for the best papers on adversarial robustness research at a workshop at ECCV, the main fall computer vision conference. The best papers are selected to have higher relevance to long-term threat models than usual adversarial robustness papers.

The Confido Institute is working on developing a user-friendly interactive app, Confido, for making forecasts and communicating beliefs and uncertainty within groups and organizations. They are also building interactive educational programs about forecasting and working with uncertainty based around this app.

This regrant will support 1.5-3 years of salary for a mathematics researcher to work with Vanessa Kosoy on the learning-theoretic AI safety agenda.

This regrant will support two AI researchers, one or two additional assistants, and a consulting firm to continue to build out and fully implement the quantitative model for how to understand risks and interventions around AI safety, expanding on their earlier research on “Modeling Transformative AI Risk.”

This regrant will support the creation of an “impact market.” The hope is to improve charity fundraising by allowing profit-motivated investors to earn returns by investing in charitable projects that are eventually deemed impactful.

This regrant will support the creation of the Swift Centre for Applied Forecasting, including salary for a director and a team of expert forecasters. They will forecast trends from Our World in Data charts, as well as other topics related to ensuring the long term future goes well, with a particular focus on explaining the “why” of forecast estimates.

We recommended a grant to support the Legal Priorities Project’s ongoing research and outreach activities. This will allow LPP to pay two new hires and to put on a summer institute for non-US law students in Oxford.

We recommended a grant to support two years of academic research on long-term economic growth.

We recommended a grant to support scholarships for talented and promising high school students to use towards educational opportunities and enrolling in a summer program.

We recommended an investment to support the development of universal CRISPR-based diagnostics, including paper-based diagnostics that can be used in developing-country settings without electricity.

We recommended a grant to support the hiring of several key staff for Dr. Kevin Esvelt’s pandemic prevention work. SecureBio is working to implement universal DNA synthesis screening, build a reliable early warning system, and coordinate the development of improved personal protective equipment and its delivery to essential workers when needed.

We recommended a grant to Cornell University to support Prof. Levine, as well as students and collaborators, to work on alignment theory research at the Cornell math department.

We recommended a grant to pay for research assistants over three years to support the work of a PhD student working on AI safety at Columbia University.

We recommended a grant to support the Institute’s research and policy engagement work on high skilled immigration, biosecurity, and pandemic prevention.

We recommended a grant to support a project aimed at training large language models to represent the probability distribution over question answers in a prediction market.

We recommended a grant to support research that we hope will be used to help strengthen the bioweapons convention and guide proactive actions to better secure those facilities or stop the dangerous work being done there.

We recommended an investment to support the creation of Pretty Good PPE that is comfortable, storable, simple, and inexpensive. PGPPE aims to provide protection that is better than disposable masks and cheaper than both hazmat suits and N95s.

This grant will support one year of operating expenses and salaries at the Legal Priorities Project, a longtermist legal research and field-building organization.

We recommended a grant to support ALTER, an academic research and advocacy organization, which hopes to investigate, demonstrate, and foster useful ways to improve the future in the short term, and to safeguard and improve the long-term trajectory of humanity. The organization's initial focus is building bridges to academia via conferences and grants to find researchers who can focus on AI safety, and on policy for reducing biorisk.

We recommended a grant to support Manifold Markets in building a charity prediction market, as an experiment for enabling effective forecasters to direct altruistic donations.

We recommended a grant to support a competition for work on Eliciting Latent Knowledge, an open problem in AI alignment, for talented high school and college students who are participating in Prometheus Science Bowl.

We recommended an investment to support Helix Nano running preclinical and Phase 1 trials of a pan-variant Covid-19 vaccine.

We recommended a grant to support Giving What We Can’s mission to create a world in which giving effectively and significantly is a cultural norm.

We recommended a grant to support research on how to fine-tune GPT-3 models to identify flaws in other fine-tuned language models' arguments for the correctness of their outputs, and to test whether these help nonexpert humans successfully judge such arguments.

We recommended a grant to support SI’s policy work with the United Nations system on the prevention of existential risks to humanity.

We recommended a grant for general support for their activities, including running conferences, supporting student groups, and maintaining online resources.

We recommended a grant to support Apollo’s work aggregating the views of academic experts in many different fields and making them freely available online.

We recommended a grant to support the development of a cheap, scalable, and decentralized platform for the rapid generation of disease-neutralizing therapeutic antibodies.

We recommended a grant to support outreach to help students to learn about career options, develop their skills, and plan their careers to work on the world’s most pressing problems.

We recommended a grant to support research on the feasibility of inactivating viruses via electromagnetic radiation.

We recommended a grant to support Lightcone’s ongoing projects including running the LessWrong forum, hosting conferences and events, and maintaining an office space for Effective Altruist organizations.

We recommended a grant to support HIA’s work encouraging professional athletes to donate more of their earnings to high impact charities and causes, and to promote a culture of giving among their fans.

We recommended a grant to support BERI in hiring a second core operations employee to contribute to BERI’s work supporting university research groups.

We recommended a grant to support a postdoc and two research assistants for Professor Braumoeller’s MESO Lab for two years to carry out research on international orders and how they affect the probability of war.

We recommended a grant to support Longview’s independent grantmaking on global priorities research, nuclear weapons policy, and other longtermist issues.

We recommended a grant to support the completion of a book which explores the nature of human values and the implications for aligning AI with human preferences.

We recommended a grant to support EffiSciences’s work promoting high impact research on global priorities (e.g. AI safety, biosecurity, and climate change) among French students and academics, and building up a community of people willing to work on important topics.

We recommended an investment to build a communication platform that provably leaks zero metadata.

We recommended a grant to support 1DS’ work on pandemic preparedness, including advocacy for advance market purchase commitments, collaboration with the UK Pandemic Ethics Accelerator on challenge studies, and advocacy with 1Day Africa and the West African Health Organization for a global pandemic insurance fund.

We recommended a grant to support the development and application of a Minimum Viable Product of a data-driven approach to improving advocacy in areas of importance to societal well-being such as immigration policy.

This grant will support a project to develop new plastic surfaces incorporating molecules that can be activated with low-energy visible light to eradicate bacteria and kill viruses continuously. If successful, this project will change how plastic surfaces are currently decontaminated.

We recommended a grant to support the identification of children in India from under-resourced areas who excel in math, science, and technology, and enable them to obtain high quality online education by digitally connecting them with mentors and teachers.

We recommended a grant to support the creation of a talent search organization which will help identify top young students around the world through a free to use website consisting of both challenging math and physics olympiad-style problems and discussion forums. Efforts will be particularly focused across India and China. These students will later be connected to support and programs so they can go on to work on the world's most pressing issues.

We recommended a grant to support the development of statewide ballot initiatives to institute approval voting. Approval voting is a simple voting method reform that lets voters select all the candidates they wish.

Longview Philanthropy and the Future Fund recommended a grant to support prizes for outstanding writing which encourages a broader public conversation around effective altruism and longtermism.

We recommended a grant to support Pathos Labs to produce a PopShift convening connecting experts on the future of technology and existential risks with television writers to inspire new ideas for their shows.