These monthly posts originated as the "Updates" section of the EA Newsletter.
You can see previous updates in our repository of past newsletters.
This month, Rob Wiblin spoke to Ajeya Cotra about worldview diversification and how big the future could be.
80,000 Hours also released two of Rob’s appearances on other podcasts: one on self-improvement and research ethics (Clearer Thinking), and the other on how he ended up the way he is (Eureka).
Finally, Michelle Hutchinson wrote three posts for the EA Forum:
- “Training Bottlenecks in EA (professional skills)”
- “Possible gaps in the EA community”
- “80,000 Hours one-on-one team plans, plus projects we’d like to see” (co-written with Niel Bowerman)
Anima International launched a program to encourage decision-makers such as politicians to be more interested in the issue of broiler chicken welfare. Improvements were proposed, including legislative solutions.
Otwarte Klatki’s plant-based team, Roślinniejemy, hosted the Best Plant-Based Products of 2020 event. Some of the awarded companies were the biggest chain stores in Poland, and the companies used social media to draw attention to the voting and the awards.
Animal Charity Evaluators
Animal Charity Evaluators (ACE) recently compiled a list of frequently asked questions about their Animal Advocacy Research Fund. The FAQ covers open science policies, project funding, and more. They also published a comparison chart of their 2020 recommended charities. Finally, ACE is asking that you help them measure their impact by reporting any of your donations that their recommendations influenced in the last year.
Animal Ethics published their plans for 2021. They intend to publish at least 70 new posts in all the languages they work in. These will include at least six reports and policy blueprints and ten research papers. They also plan to publish up to 70 new audio recordings in their chosen languages. The 2021 plans are also available in Spanish, Portuguese, and German.
In January, Animal Ethics also published “Establishing a field in natural sciences: three case studies.” The study explores how academic fields related to animals are formed, and how these fields can be helpful for those working in defense of animals, particularly for animals living in the wild. The report contains interviews with leading experts in different fields and investigates the available literature about the fields’ origins.
Berkeley Existential Risk Initiative
BERI is hiring a personal assistant for Professor Stuart Russell, director of the Center for Human-Compatible Artificial Intelligence. You can read more and apply here.
Centre for Effective Altruism
Ben West published three new posts on retention in the EA community. He covers the reasons people choose to leave or stay, project ideas for increasing retention, and how EA’s retention rates compare to those of other social groups.
Center for Human-Compatible AI
Stuart Russell debated Melanie Mitchell, a computer science professor at the Santa Fe Institute, in the latest episode of The Munk Debates, a debate series on major policy issues. In the episode, titled “The Rise of Thinking Machines,” Prof. Russell argued in favor of a resolution that “the quest for true AI is one of the great existential risks of our time”. Listen to the debate here or on any podcast service.
Rachel Freedman spoke to the Berkeley High School STEMinist Club on the topic of human-compatible AI research.
The International Conference on Learning Representations (ICLR) 2021 accepted the following papers from CHAI researchers:
- Adam Gleave, Michael Dennis, Shane Legg, Stuart Russell, Jan Leike: “Quantifying Differences in Reward Functions” (spotlight paper)
- Dan Hendrycks, Collin Burns, Steven Basart, Andy Zou, Mantas Mazeika, Dawn Song, Jacob Steinhardt: “Measuring Massive Multitask Language Understanding”
- Dan Hendrycks, Collin Burns, Steven Basart, Andrew Critch, Jerry Li, Dawn Song, Jacob Steinhardt: “Aligning AI With Shared Human Values”
- David Lindner, Rohin Shah, Pieter Abbeel, Anca Dragan: “Learning What To Do by Simulating the Past”
“Choice Set Misspecification in Reward Inference” by Rachel Freedman, Rohin Shah, and Anca Dragan received the Best Paper Award at the IJCAI 2020 AISafety workshop.
Tom Krendl Gilbert published the white paper “Mapping the Political Economy of Reinforcement Learning Systems: The Case of Autonomous Vehicles” on the Simons Institute website.
Center on Long-Term Risk
Daniel Kokotajlo published “Birds, Brains, Planes, and AI: Against Appeals to the Complexity/Mysteriousness/Efficiency of the Brain” (Alignment Forum).
People sometimes argue that superhuman AGI will take a long time to develop by appealing to the complexity, mysteriousness, and efficiency of the human brain compared to modern AI systems. Daniel claims that these arguments are fundamentally flawed, illustrating his points with the case of powered flight vs. birds. This historical precedent is also weak evidence that we should look to things like neuron count to predict when AI capabilities will equal those of humans, since flight was achieved when motors achieved similar power-to-weight ratio as bird muscle.
Centre for the Study of Existential Risk
CSER has published “80 Questions for UK Biosecurity.” The paper provides a methodology and research agenda that can also guide the evaluation of biosecurity in other countries.
Matthijs Maas and Charlotte Stix have published a new paper in AI and Society: “Bridging the gap: the case for an ‘Incompletely Theorized Agreement’ on AI policy.” They propose that in certain areas, scholars working on challenges from near-term and long-term AI can converge and cooperate on mutually beneficial AI policy projects, while maintaining their distinct perspectives.
CSER supported the work of the Science 20 (S20) at the G20 by collaborating in the production of the report “Foresight: Science for Navigating Critical Transitions.”
Esme Booth has joined the CSER team as a Project Coordinator. She has a background in project management and communications within academia.
Luke Kemp has signed a contract to publish a forthcoming book, Downfall: The Eight Forces that End Civilisations, with Penguin's Viking imprint.
Charity Entrepreneurship (CE) recently hosted an online event, “Impactful opportunities around and adjacent to charity entrepreneurship.” In case you missed it, the recording is now available. The CE community also shared their tips on early-stage hiring for new nonprofit start-ups in a roundtable discussion.
Effective Altruism Funds
The Animal Welfare Fund, the Long-Term Future Fund, and the EA Infrastructure Fund are calling for funding applications. If you have a project you think will improve the world, and it seems like a good fit for one of the funds, apply by 7 March (11:59 p.m. PST). Apply here, or see further information.
The Long-Term Future Fund and EA Infrastructure Fund now support anonymized grants; if you prefer not to have your name listed in the public payout report, EA Funds is still interested in funding you.
Faunalytics released a new analysis of Twitter trends in animal protection. They examined how public engagement with topics including animal-friendly diets, factory farming, effective altruism, and other advocacy issues changed over the course of a year. The results are available here, and the raw data is available by request.
They’ve also updated their Research Advice page with information on conducting literature reviews, designing a study, collecting data, and analyzing results. Finally, they produced a new blog on identifying and mitigating burnout, and added several new study summaries to their library on topics including global insect farming, gaps in international wildlife trade regulation, and consumer acceptance of cultured meat.
Fish Welfare Initiative
Fish Welfare Initiative secured their first commitment to improve fish welfare. The organization that committed, Gramodaya Trust, works in India with several hundred fish farmers. FWI estimates that this will improve the lives of several million fish over the coming few years.
FWI’s Fish Welfare Specialist, Marco Cerqueira, recently gave an interview to The Fish Site, a leading industry knowledge hub.
FWI also recently gained their 501(c)(3) status, which will allow them to operate more independently and reduce operational costs.
Lastly, they are still hiring for both a Generalist Intern and Communications Intern. All are welcome to apply for the communications role, but generalist candidates must “[be] from or have cultural fluency with a country in South, East, [or] Southeast Asia”.
Future of Humanity Institute
In January, Ryan Carey co-authored “Agent Incentives: A Causal Perspective,” which has been accepted to the AAAI 2021 Conference. The authors of the paper also launched a new website for the Causal Incentives Working Group.
Lewis Hammond co-authored “Equilibrium Refinements for Multi-Agent Influence Diagrams: Theory and Practice,” which has been accepted to the 20th International Conference on Autonomous Agents and Multiagent Systems.
Lukas Finnveden wrote a post for the Alignment Forum titled “Extrapolating GPT-N performance,” and Stuart Armstrong also contributed two posts: “Syntax, semantics, and symbol grounding, simplified” and “Model splintering: moving from one imperfect model to another.”
Anders Sandberg co-authored “What is the Upper Limit of Value?” and Markus Anderljung and Allan Dafoe co-authored “The Immigration Preferences of Top AI Researchers: New Survey Evidence.”
Jonas Sandbrink and Joshua Monrad, two of FHI’s 2020 Summer Research Fellows, published “Promoting versatile vaccine development for emerging pandemics” in NPJ Vaccines. Jonas also published “Safety and security concerns regarding transmissible vaccines” in Nature Ecology & Evolution.
Future of Life Institute
- Nominations for the 2021 Future of Life Award are now open. This prize is awarded to an individual who, without having received much recognition at the time, has helped make today dramatically better than it otherwise would have been. Previous winners include Bill Foege and Viktor Zhdanov, who made critical contributions towards the eradication of smallpox, and Matthew Meselson, the driving force behind the 1972 Biological Weapons Convention. There is a $3,000 prize for the person who nominates the winner!
- To commemorate the United Nations’ Treaty on the Prohibition of Nuclear Weapons, FLI launched nuclearweapons.info. This site represents a collaboration of two dozen organisations and is intended to serve as a single location where you can go to learn about the threat of nuclear weapons, the organizations involved in ending this threat, and easy ways to get involved.
Donors in the Netherlands can now make tax-deductible gifts through GiveWell. More details can be found here.
GiveWell also published details on their November 2020 giving recommendation to Open Philanthropy.
GiveWell is currently seeking researchers to identify, analyze, and compare the giving opportunities that can most cost-effectively save or improve the lives of the global poor:
- Senior Research Associate: You have more than six years of relevant work or educational experience, often including a master's degree or PhD.
- Senior Researcher: You have over a decade of relevant work experience, often involving both a PhD and a few years of work experience, or a master's degree and many years of work experience.
GiveWell is also seeking a Philanthropy Advisor to build long-term relationships with GiveWell supporters.
Giving What We Can
2020 was Giving What We Can’s biggest year on record, and included many significant milestones. There were 1,787 new pledges (1,008 members of The Pledge and 779 of Try Giving) and member donations of over $20 million. There are now more than 5,000 total members of The Pledge, who have reported over $200 million in donations to high-impact charities and pledged over $2 billion in total donations over the course of their lives.
New members came from a wide range of sources. The most common were 80,000 Hours, various podcasts, and referrals by friends and family (a great example of how vital member advocacy is). GWWC’s top-viewed resources included the How Rich Am I Calculator, Giving Recommendations and the classic essay “Famine, Affluence, and Morality” by Peter Singer. Top new blog posts included a video message to our 5,000 members, a member profile of Arvind Raghavan, and the announcement of our Company Pledge.
Global Catastrophic Risk Institute
GCRI published “2020 Survey of Artificial General Intelligence Projects for Ethics, Risk, and Policy.” The paper is an updated version of GCRI’s 2017 survey of the same name. GCRI’s 2020 survey uses improved methodology to find and accurately characterize 72 different projects operating in 37 countries, and shows how the AGI R&D landscape changed between 2017 and 2020.
In addition, GCRI is pleased to announce that it has received $209,000 in general support from Jaan Tallinn through the Survival and Flourishing Fund. We are grateful for this generous support of our work as we develop ways to confront humanity’s gravest threats.
Global Priorities Institute
GPI has launched a new essay prize for global priorities research. Graduate students currently pursuing master's-level courses in philosophy on topics in global priorities research are eligible. The winning entry will be published as a working paper on GPI’s website, and the winner will receive a prize of £1,000. The deadline for submissions is 1 July 2021.
In addition, GPI published their latest annual report, which covers their work during the last academic year.
The Humane League
Last month, THL launched their 2021 Cage-Free Eggsposé: a public list of the companies that pledged to eliminate cages from their egg supply chain but didn’t keep to their pledges. The release of the Eggsposé coincided with the launch of a grassroots campaign blitz against these companies. Pressure from activists has already led to four of the thirteen reporting their progress, including Arby’s and Lucky’s Market publicly reporting they are already 100% cage-free.
THL UK is advocating for farmed fish, including conducting message testing and lobbying for slaughter reform. THL UK’s campaigners discussed the initiative in a recording, viewable here.
Kalista Barter, VP of Development, outlined THL’s strategy for 2021.
IDinsight recently published a working paper on how to design decision-focused impact evaluations for policymakers. The organization also recently published research on how cash transfers impacted refugees’ food security in a settlement in Uganda.
IDinsight CEO Ruth Levine, GiveWell Managing Director Neil Buddy Shah (former IDinsight CEO and Founding Partner), BRAC Executive Director Asif Saleh, and Co-Impact’s Rakesh Rajani spoke about evidence-based philanthropy at the Yale Philanthropy Conference on 12 February.
IDinsight is hiring for many roles around the world.
Open Philanthropy announced grants and investments including $5.5M to VasoRX to support work testing a new therapy for vascular diseases, $2M to Binx to support work on a COVID-19 diagnostic platform, $1.2M to Compassion in World Farming to support work to advance farm animal welfare in Asia, and $1M to the International Refugee Assistance Project for general support. The organization also published a blog post explaining their approach to recruiting a strong team.
Ought is building research assistant features into Elicit. Researchers teach Elicit how to perform a task by providing a few examples. Elicit then uses GPT-3 to scale that task to thousands of publications, questions, and data sources. Demos of these features are here. If you’re interested in beta testing and creating new research tasks, sign up at ide.elicit.org!
Qualia Research Institute
QRI released several announcements:
- They’ve chosen a new Executive Director: Andrew Zuckerman
- Their new Board of Advisors includes Wojciech Zaremba, Dr. Robin Carhart-Harris, David Pearce, Scott Alexander, and Dr. Shamil Chandaria.
- History and 2021 Strategy: this post outlines QRI’s mission, history, recent achievements, organizational changes, and goals for 2021. In 2021, they plan to empirically explore the Symmetry Theory of Valence (STV), continue neurotech development, publish a paper that argues for STV on theoretical and empirical grounds, release open-source neuroimaging software, expand their psychophysics toolkit, and work on generalizing their CSHW (connectome-specific harmonic waves) framework into a scale-free form.
- 2021 fundraising campaign: QRI is raising $1.5 million to hire three new people (two engineers and a computational neuroscientist), pay their existing team, invest in computing power, and, post-COVID, move to a physical location. The campaign is detailed in the History and 2021 Strategy post. If you’d like to contribute, you can donate on their website.
Wild Animal Initiative
Wild Animal Initiative held their annual strategy retreat. Staff met virtually to discuss the organization’s long-term goals, revise their strategic plan, and set goals for 2021.
Former Staff Researcher Jane Capozzelli moved on to her new position as a Partner Avian Biologist with the West Virginia Division of Natural Resources. In an interview on Wild Animal Initiative’s blog, she reflects on her time at WAI and how she came to devote her career to helping wild animals.
Add your own update
If your organization isn't shown here, you can provide an update in a comment.
You can also email me if you'd like to be one of the organizations I ask for updates each month. (I may not accept all such requests. Whether I include an org depends on its size, age, focus, track record, etc.)