Awards for the Future Fund’s Project Ideas Competition

By Nick_Beckstead @ 2022-03-24T18:14 (+151)

This post announces the winners of the Future Fund’s Project Ideas Competition, and reflects on the process of running the competition.

We got an overwhelming response to the competition, receiving close to 1000 submissions.  The blog post received more comments than any other post on the Forum; in fact, more than double the second-most-commented post. We were thrilled at the level of excitement that the competition generated. So that we can appropriately reward the submissions, we’ve decided to include a category of “honorable mentions”: these will not go onto the website, but will receive an award of $1000. 

We will be contacting the winners individually about how to receive their prizes.

Winners

The winners - which will soon go up on our project ideas page (often in modified form) - are as follows:

gavintaylor - Infrastructure to support independent researchers

Epistemic Institutions, Empowering Exceptional People  

The EA and Longtermist communities appear to contain a relatively large proportion of independent researchers compared to traditional academia. While working independently can provide the freedom to address impactful topics by liberating researchers from the perverse incentives, bureaucracy, and other constraints imposed on academics, the lack of institutional support can impose other difficulties that range from routine (e.g. difficulties accessing pay-walled publications) to restrictive (e.g. lack of mentorship, limited opportunities for professional development). Virtual independent scholarship institutes have recently emerged to provide institutional support (e.g. affiliation for submitting journal articles, grant management) for academic researchers working independently. We expect that facilitating additional and more productive independent EA and Longtermist research will increase the demographic diversity and expand the geographical inclusivity of these communities of researchers. Initially, we would like to determine the main needs and limitations independent researchers in these areas face and then support the creation of a virtual institute focussed on addressing those points.

Konstantin Pilz - EA content translation service

Effective Altruism, Movement-Building

EA-related texts are often using academic language needed to convey complex concepts. For non-native speakers reading and understanding those texts takes a lot more time than reading about the same topic in their native language would. Furthermore, today many educated people in important positions, especially in non-western countries, do not speak or only poorly speak English. (This is likely part of the reason that EA currently mainly exists in English speaking countries and almost exclusively consists of people speaking English well.)

To make EA widely known and easy to understand there needs to be a translation service enabling e.g. 80k, important Forum posts or the Precipice to be read in different languages. This would not only make EA easier to understand - and thus spread ideas further - but also likely increase epistemic diversity of the community by making EA more international.

Mackenzie Arnold - A regulatory failsafe for catastrophic or existential biorisks

Biorisk and Recovery from Catastrophes

Currently, many government regulators (like the FDA in the US) apply a static set of criteria when evaluating countermeasures used to fight disease or other public harms. While these criteria may operate relatively well during normal times, during catastrophic events, they would likely impose overly cautious limitations on response efforts and, in some cases, may even prohibit the development or deployment of countermeasures with relatively minor risk profiles relative to the threat at hand. To avoid such a system, we would be interested in supporting work that aims to research, develop, or advocate for (through policy or legal challenges) alternative regulatory structures that would better accommodate the needs of catastrophic risks scenarios. 

Thanks to Kyle Fish for discussing this and related ideas over the past year.

Marc-Everin Carauleanu - Datasets for AI alignment research

Artificial Intelligence

The success of Machine Learning experiments relies heavily on the quality and quantity of training data which is oftentimes difficult and expensive to obtain. We would like to see an organization that has the infrastructure and capacity to provide training data for any promising AI Alignment research proposal. This could aid the development of alignment-relevant metrics in line with Open Philanthropy's 'Measuring and forecasting risks' research direction as well as potentially incentivize ML researchers to focus on alignment work as a key bottleneck - training data - will be taken care of. We would hope that datasets developed by this organization have the potential to transform AI Alignment research similarly to how ImageNet accelerated Computer Vision research. 

Elizabeth Barnes - High-quality human data

Artificial Intelligence

Most proposals for aligning advanced AI require collecting high-quality human data on complex tasks such as evaluating whether a critique of an argument was good, breaking a difficult question into easier subquestions, or examining the outputs of interpretability tools. Collecting high-quality human data is also necessary for many current alignment research projects. 

We’d like to see a human data startup that prioritizes data quality over financial cost. It would follow complex instructions, ensure high data quality and reliability, and operate with a fast feedback loop that’s optimized for researchers’ workflow. Having access to this service would make it quicker and easier for safety teams to iterate on different alignment approaches.

Some alignment research teams currently manage their own contractors because existing services (such as surge.ai and scale.ai) don’t fully address their needs; a competent human data startup could free up considerable amounts of time for top researchers.

Such an organization could also practice and build capacity for things that might be needed at ‘crunch time’ – i.e., rapidly producing moderately large amounts of human data, or checking a large volume of output from interpretability tools or adversarial probes with very high reliability. 

The market for high-quality data will likely grow – as AI labs train increasingly large models at a high compute cost, they will become more willing to pay for data. As models become more competent, data needs to be more sophisticated or higher-quality to actually improve model performance. 

Making it less annoying for researchers to gather high-quality human data relative to using more compute would incentivize the entire field towards doing work that’s more helpful for alignment, e.g., improving products by making them more aligned rather than by using more compute.

[Thanks to Jonas V for writing a bunch of this comment for me]

[Note from Nick: we'll probably add just one of the above two ideas to our site, or some amalgamation.]

Mark Xu - Detailed stories about the future

Artificial Intelligence

We're interested to see stories about how the present evolves into the future that are as specific and realistic as possible. Such stories should be set in a world that is "on trend" with respect to technological development and aim to consider realistic sets of technologies coexisting in a global economy. We think such stories might help make it easier to feel, rather than just abstractly understand, that this might be the most important century.

We will also award a $5000 prize to Fin Moorhouse, for his list of EA projects. The ideas are not of the right format to immediately go on the website, and we think the ideas make more sense to be directed specifically at the EA community than to go on the Future Fund website. But we thought that the write-up was impressive, and we want to reward it.

Honorable mentions

We are also awarding $1000 "honorable mention" prizes for the following suggestions:

agnode - SEP for every subject

Epistemic institutions

Create free online encyclopedias for every academic subject (or those most relevant to longtermism) written by experts and regularly updated. Despite the Stanford Encyclopedia of Philosophy being widely-known and well-loved there are few examples from other subjects. Often academic encyclopedias are both behind institutional paywalls and not accessible on sci-hub (e.g. https://oxfordre.com/). This would provide decision makers and the public with better access to academic views on a variety of topics.

JacksonWagner - Resilient ways to archive valuable technical / cultural / ecological information

Biorisk and recovery from catastrophe

In ancient Sumeria, clay tablets recording ordinary market transactions were considered disposable. But today's much larger and wealthier civilization considers them priceless for the historical insight they offer. By the same logic, if human civilization millennia from now becomes a flourishing utopia, they'll probably wish that modern-day civilization had done a better job at resiliently preserving valuable information. For example, over the past 120 years, around 1 vertebrate species has gone extinct each year, meaning we permanently lose the unique genetic info that arose in that species through millions of years of evolution.

There are many existing projects in this space -- like the internet archive, museums storing cultural artifacts, and efforts to protect endangered species. But almost none of these projects are designed robustly enough to last many centuries with the long-term future in mind.  Museums can burn down, modern digital storage technologies like CDs and flash memory aren't designed to last for centuries, and many critically endangered species (such as those which are "extinct in the wild" but survive in captivity) would likely go extinct if their precarious life-support breeding programs ever lost funding or were disrupted by war/disaster/etc. We're potentially interested in funding new, resilient approaches to storing valuable information, including the DNA sequences of living creatures.

JanBrauner - Cognitive enhancement research and development (nootropics, devices, ...)

Values and Reflective Processes, Economic Growth

Improving people's ability to think has many positive effects on innovation, reflection, and potentially individual happiness. We'd like to see more rigorous research on nootropics, devices that improve cognitive performance, and similar fields. This could target any aspect of thinking ability---such as long/short term memory, abstract reasoning, creativity---and any stage of the research and development pipeline---from wet lab research or engineering over testing in humans to product development.

Jared Mueller - "Building vibrant EA academic communities in Africa, Asia, and Latin America"

The largest EA university communities are concentrated in Europe and North America. We would like to support the emergence of more broad-based and vibrant EA academic communities across Africa, Asia, and Latin America. This will enhance the cultural diversity of EA, and broaden the supply of students and faculty tackling the most pressing problems. The Universities of Ibadan and São Paulo, India's IITs and IIMs, and the National Autonomous University of Mexico are only a few examples of campuses where we would be eager to see thriving EA intellectual communities.

Kat Woods - Translate EA content at scale

Reach More Potential EAs in Non-English Languages

Problem: Lots of potential EAs don’t speak English, but most EA content hasn’t been translated.

Solution: Pay people to translate the top EA content of all time into the most popular languages, then promote it to the relevant language communities.

Keiran Harris - Betting exchange focused on debates between individuals 

Bryan Caplan argues that “bets are one of the best ways to (a) turn vague verbiage into precise statements, and (b) discover the extent of genuine disagreement about such precise statements.” But it isn't easy to set up bets with people you don’t already trust, so many potential bets don’t get made. If instead one person could say, “I just set up a bet on X.com on our point of disagreement, here’s the link if you want to accept” — many more bets might get made. The site could also keep a public record of bettor’s long-run track records — which Caplan thinks is one of the best ways to assess thinkers’ credibility.

Lennart Stern and John Halstead - The Mission Innovation Initiative

The Mission Innovation Initiative tracks public spending on different categories of clean energy RD&D. For each category, a fund could be created with the mandate to use its budget to maximize the rate of progress on the specific type of clean energy RD&D through incentive payments made to countries. When a global fund directly spends money on projects in areas in which the government already spends substantial amounts, there is the possibility that the governments’ spending will be crowded out. Theoretically, there are therefore advantages to incentivising countries to spend more. Stern (2021) finds that 1 billion dollars spent through the optimal such mechanism causes an increase in aggregate clean energy RD&D of 5 billion.

Fossil fuels combustion can be curbed through restricting supply, restricting demand and expanding substitutes. Here is an analysis suggesting that it is currently best for new global funds to focus on substitute expansion.

New scalable Global Public Good Institutions like the one proposed could be funded through novel mechanisms like the MGF mechanism proposed here.

MaxG - DIY decentralized nucleic acid observatory

Biorisk and Recovery from Catastrophes

As part of the larger effort of building an early detection center for novel pathogens, a smaller self-sustaining version is needed for remote locations. The ideal early-detection center would not only have surveillance stations in the largest hubs and airports of the world, but also in as many medium sized ones as possible. For this it is necessary to provide a ready-made, small and transportable product which allows meta-genomic surveillance of wastewater or air ventilation. One solution would be designing a workflow utilizing the easily scalable and portable technology of nanopore sequencing and combining it with a workflow to extract nucleic acids from wastewater. The sharing of instructions on how to build and use this method could lead to a "do it yourself" (DIY) and decentralized version of a nucleic acid observatory. Instead of staffing a whole lab at a central location, it would be possible to only have one or two personnel in key locations who use this product to sequence samples directly and only transmit the data to the larger surveillance effort.

Nicholas Schiefer - Special economic zones near the United States

There are tons of talented people outside of the United States who would like to live there and participate in its dynamic economy and remarkable institutions. However, US immigration law makes this very difficult, even for very talented people. Meanwhile, high productivity regions within the United States tend to have extremely high costs of living, primarily due to shortages of housing.

We’d love to solve both of these problems by working with a place near the United States (such as a Caribbean or Atlantic Island, or perhaps a Canadian province or Mexican State) to set up a “special economic zone” (SEZ). An SEZ would target explicit harmonization with American law on matters related to doing business and research, making it easy for American organizations to hire people or set up offices there. A SEZ would have extremely non-restrictive immigration policies, allowing people to move there from basically anywhere in the world. They would also set policy carefully so that basic goods like housing remain affordable (for example, by forbidding restrictions on the height of buildings).

Pablo - Retrospective grant evaluations

Research That Can Help Us Improve

EA funders allocate over a hundred million dollars per year to longtermist causes, but a very small fraction of this money is spent evaluating past grantmaking decisions. We are excited to fund efforts to conduct retrospective evaluations to examine which of these decisions have stood the test of time. We hope that these evaluations will help us better score a grantmaker’s track record and generally make grantmaking more meritocratic and, in turn, more effective. We are interested in funding evaluations not just of our own grantmaking decisions (including decisions by regrantors in our regranting program), but also of decisions made by other grantmaking organizations in the longtermist EA community.

RoryFenton - Campaign to eliminate lead globally

Economic Growth

Lead exposure limits IQ, takes over 1 million lives every year and costs Africa alone $130 billion annually, 4% of GDP: an extraordinary limit on human potential. Most lead exposure is through paint in buildings and toys. The US banned lead paint in 1978 but 60% of countries still permit it. We would like to see ideas for a global policy campaign, perhaps similar to Bloomberg’s $1 billion tobacco advocacy campaign (estimated to have saved ~30 million lives), to push for regulations and industry monitoring.

Rumtin Sepasspour - Existential risk whistleblowing mechanism

Say you’re an employee of an organization building AGI without safety measures or a government official that sees your leadership massively increasing risk, how do you raise the alarm? You risk personal, career or financial costs if you try to push back on these concerns internally. There needs to be some incentives, such as financial compensation, to bring these concerns to public attention. It may also require legal support or protection, and a clear avenue for making an impact with your revelations. This may be particularly important in authoritarian countries, where the financial compensation may be extremely high in comparison to a salary, and policy change is unlikely from inside the system. We need more mechanisms that allow people to raise these concerns and reinforce norms around existential security.

RyanCarey - EA Coworking Spaces at Scale

Effective Altruism

The EA community has created several great coworking spaces, but mostly in an ad hoc way, with large overheads. Instead, a standard EA office could be created in upto 100 towns and cities. Companies,  community organizers, and individuals  working full-time on EA projects would be awarded a membership that allows them to use these offices in any city. Members gain from being able to work more flexibly, in collaboration with people with similar interests (this especially helps independent researchers with motivation). EA organizations benefit from decreased need to do office management (which can be done centrally without special EA expertise). EA community organizers gain easier access to an event space and standard resources, such as a library, and hotdesking space, and some access to the expertise of others using the office.

Toby Shevlane - Tools that facilitate structured access to powerful AI systems

Structured access is an approach to sharing AI models that allows people to use and study the model, but only within a structure that prevents misuse and undesired information leaks. There are early examples of structured access (e.g. OpenAI’s GPT-3 API) but the paradigm has not yet reached maturity. It would likely be possible to give external researchers greater flexibility to study models (including for the purposes of safety) even without significantly increasing the likelihood of misuse and unwanted proliferation. Also, tools for granting structured access to AI models are not widely available, which is a barrier for labs adopting a structured access approach. We would be excited to fund a fellowship for technical researchers to build open source tools for structured access. This could be in collaboration with OpenMined, an existing open source community focussed on similar topics.

Zdgroff - Advocacy for digital minds

Artificial Intelligence, Values and Reflective Processes, Effective Altruism

Digital sentience is likely to be widespread in the most important future scenarios. It may be possible to shape the development and deployment of artificially sentient beings in various ways, e.g. through corporate outreach and lobbying. For example, constitutions can be drafted or revised to grant personhood on the basis of sentience; corporate charters can include responsibilities to sentient subroutines; and laws regarding safe artificial intelligence can be tailored to consider the interests of a sentient system. We would like to see an organization dedicated to identifying and pursuing opportunities to protect the interests of digital minds. There could be one or multiple organizations. We expect foundational research to be crucial here; a successful effort would hinge on thorough research into potential policies and the best ways of identifying digital suffering.

Some reflections

Here are some quick notes on how this project went relative to our expectations:

• Our initial best guess was that we'd add 5-10 project ideas to our website, and so that aspect of the final outcome matched our expectations.
• We got about twice as many proposals as we expected.
• It was notable that we didn’t see any idea that felt exceptional enough to go over the $5000 payout. Most of the ideas we liked were familiar to us and the main value-add was either reminding us of a potential project that we’d forgotten about or a clear written explanation. This was somewhat surprising to us.

The project was overall a pretty significant amount of work, and we are undecided about how often to use this kind of approach in the future. Still, we are overall happy we did the experiment, and we were happy about the excitement it created around generating concrete project ideas.

Here are some quick notes on what kinds of project ideas we found more or less helpful:

• We found project ideas much more useful when they were more concrete. There’s a huge difference in usefulness to us between “We should hire someone to run X to help with Y / we should do research on X” and “This particular line of attack would be helpful.”
• We found project ideas more helpful when there was a clear reason that there was a market failure that was being corrected. For example, a common category was “X for EAs” where X is some service that seems like it would be best provided by normal market mechanisms.
• We were excited about concrete projects that weren't just about promoting EA or longtermism. We really want the EA community to achieve concrete wins and launch ambitious and inspiring object-level projects to improve the long-term future. We think that will help the world directly, improve our culture, and also help with recruitment in the long run.