Thanks to everyone who participated in ACX Grants, whether as an applicant, an evaluator, or a funder.
We received 654 applications this year, and were able to fund 42. To the other 612: sorry! Many of you had great ideas that we couldn’t fund for contingent reasons - sometimes because we couldn’t evaluate them at the level of depth it would have taken to feel comfortable supporting them, or because we had complicated conflicts of interest, or just because we didn’t have enough money. Some of you had ideas that were good but not a match for our particular grantmaking philosophy. Finally, a few of you were suffering from LLM psychosis. Please get help.
Of the 42 grantees, 40 have answered our email asking for confirmation that they still want the grant. I’m still waiting for confirmation emails from Lewis Wall and Nishank B. If you’re reading this and don’t think you got a confirmation email, check your spam folder. If it’s not in your spam folder, email me at scott@slatestarcodex.com. If you can’t reach me or I don’t respond, DM me on Substack or Twitter. I’ll give you until November 1 to get in touch, after which point the grant will be withdrawn. There are also a few projects so deep in stealth I don’t have permission to share their existence; I will mention these as they become public.
More information, and the all-important thanks to contributors, are after the list, which is:
Kasey Markel, $10K, for genetically engineered corn. Kasey and his team at Semilla Nueva use prime editing, a new genetic technology, to create corn which is rich in zinc, iron, essential amino acids, and other nutrients frequently deficient in corn-heavy poor country diets. Our grant helps fund greenhouse space, enzymes, DNA synthesis, and scientist time, and will let them expand faster into new regions that require corn with different genetic backgrounds.
Maximillian Seunik, $50K, for Screwworm Free Future. The screwworm is a nasty flesh-eating worm that infests cattle and occasionally humans. It was laboriously eliminated from the US in the 1960s, from Mexico and Central America in the 90s, and finally fought to a standstill along the defensible chokepoint of the Panama isthmus in 2006. Since then, the US has regularly dropped sterile male screwworms over Panama; these distract the females and prevent them from advancing back north. During COVID, the parasite breached the barrier; it’s now back as far as Mexico, and likely to re-enter the US soon. SFF wants to encourage the development and testing of genetic biocontrol approaches, alongside other technology, to rapidly suppress screwworm populations. If these techniques work in screwworms, they could later be applied to mosquitoes, ticks, and other pests.
Markus Englund, $50K, for software to detect data fabrication. This kind of thing is a perennial ACX Grants favorite, and we don’t always expect it to go anywhere, but Markus got our attention by saying that he’s already built the tool, already scanned 92 published papers, and found “irregularities” in five of them, inspiring two corrigenda and one likely upcoming retraction. Five out of ninety-two is a crazy result, and we’re almost scared to see what happens when he applies his program to a further 20,000 papers, which is the amount that our grant will be paying for. If you’re interested in helping verify cases of suspected data fabrication and presenting the evidence in Pubpeer comments or emails to journal editors, please contact Markus at markus@englund.dev, especially if you have solid knowledge of statistics or biology.
Micaella Rogers and Tom Daniels, $50K, for lead-acid battery recycling. Unsafe lead-acid battery recycling is a major contributor to global lead burden; it’s hard to figure out how literally and causally to take the highest estimates of damage, but they suggest up to 350,000 deaths per year and $170 billion in lost productivity. Some governments have curtailed this problem by making customers pay a deposit along with a new battery, which they get back when they return the battery to a safe recycling facility. Micaella and Tom’s organization wants to advise the Philippines government on how to do the same.
Aaron Silverbook, $5K, for approximately five thousand novels about AI going well. This one requires some background: critics claim that since AI absorbs text as training data and then predicts its completion, talking about dangerous AI too much might “hyperstition” it into existence. Along with the rest of the AI Futures Project, I wrote a skeptical blog post, which ended by asking - if this were true, it would be great, right? You could just write a few thousand books about AI behaving well, and alignment would be solved! At the time, I thought I was joking. Enter Aaron, who you may remember from his previous adventures in mad dental science. He and a cofounder have been working on an “AI fiction publishing house” that considers itself state-of-the-art in producing slightly-less-sloplike AI slop than usual. They offered to literally produce several thousand book-length stories about AI behaving well and ushering in utopia, on the off chance that this helps. Our grant will pay for compute. We’re still working on how to get this included in training corpuses. He would appreciate any plot ideas you could give him to use as prompts.
Charlie Mothrop, $5K, for “normie-friendly prediction market interfaces”. Charlie has already made some tools for visualizing Manifold and Polymarket results; for example, a bot that tweets sudden dramatic changes on important Manifold questions.
Ben Engebreth, $6K, for a new asteroid-hunting algorithm. Modern telescopes produce massive databases of how the sky looks at different times. Ben has developed an improved algorithm for searching these databases, linking detections in different images, and determining whether the detections match the profile of a previously-undiscovered asteroid. He wants money to buy enough compute to run his algorithm on the Rubin Observatory’s Legacy Survey of Space and Time dataset.
Lewis Wall, $50K, for therapeutic food in Ethiopia. After years of drought, war, and locusts, the Tigray region of Ethiopia is experiencing a major famine. Lewis and the Fewsi Foundation will produce a special peanut butter optimized to relieve the worst effects of childhood malnutrition. This grant will fund a giant commercial mixer to help produce the peanut butter, plus some of the raw material and distribution cost.
Daniela Shuman, $100K, to improve eligibility for organ donation. Many people want to donate an organ during their lifetime, but are turned away for minor health problems (e.g. being overweight, being a smoker). Daniela’s org, Project Donor, gives these people free high-quality medical assistance to solve their problems (e.g. lose weight, quit smoking), then encourages them to reapply. They report having caused >100 successful donations so far, but are growing fast and think there’s a “market” to enable as many as 2,000 extra transplants per year. I was excited by them not only because of my own frustrating experience with organ donation, but because they claim incredible cost-effectiveness numbers - maybe as little as $2,500 per life saved.
David Rozado, $50K, to study truth-seeking and bias in LLMs. Suppose you ask a chatbot about minimum wages, and it summarizes economic research on the topic. Or suppose it’s 2030, GPT-7 has outpaced human economists, and you want it to do original analysis. How can you be sure that it’s not falling victim to the same political biases that might plague the rest of us? Professor Rozado studies this question in depth, working on tools that measure bias (for example, whether the AI will evaluate study methodologies consistently when the results favor different political views) and trying to determine what interventions (prompts, fine-tuning, etc) best ensure AI neutrality. Philip Tetlock, of superforecasting fame, will assist with this research.
Adam Morris, $15K, to train LLMs to honestly report their internal decision processes via introspection. Conventional wisdom says AIs can’t introspect - they’re not even consistently aware they’re chatbots unless you prompt them to remember. But Adam and his collaborators have found some glimmers of surprisingly good introspective ability into decision-making processes - for example, ability to explain how past fine-tuning affects the relative values of different goods - and has some evidence that this can improve with training. He wants to create an introspection benchmark, and to see what happens when you train AIs to succeed on that benchmark. This could supplement other forms of interpretability, improve chain of thought faithfulness, and help us answer questions about AI consciousness. Adam is excited to chat with potential collaborators who have experience in technical AI safety work (especially in interpretability, CoT faithfulness, and fine-tuning frontier open models); reach out to him at thatadammorris@gmail.com.
Alexander (Olek) Pisera, $50K, for yeast-based manufacturing. Producing “biologics” - protein-based drugs like insulin or monoclonal antibodies - is often very expensive. One technique is to genetically engineer yeast to secrete the protein you want, but the yeast doesn’t always cooperate, and yields can be low. Alexander is building a platform that automates the evolution of output-increasing genes, eventually resulting in strains better optimized for this sort of production. If this works, it could help poor countries do their own biologics manufacturing, bypassing expensive middlemen and tricky logistics.
Nino O’Shea-Nejad, $5K, to investigate electrical stunning in shrimp and other crustaceans. Shrimp welfare’s inherent weirdness turned it first into a meme, then a celebrity EA cause, and finally a serious field of charity. The idea that stunning shrimp before killing them improves welfare is intuitively appealing, but the evidence base remains limited. Nino will review the scientific literature across decapod crustaceans, and identify what future research would help determine whether electrical stunning reliably renders them insensible.
David Carel, $150K, to help put air purifiers in schools. Pure air is an easy sell, but an increasing body of research suggests it may have unexpected advantages, including raising test scores in classrooms. This might just be because students with fewer respiratory diseases take fewer absences, or there might be more interesting connections between air pollution, respiratory health, focus, and achievement. Many schools bought air purifiers during COVID but forgot about them afterwards, or turned them off because they were too noisy; now they languish in closets, fully functional but unused. David wants to lobby schools to use the devices they have, and to develop quieter devices that are better suited for classrooms. If you’re a school, potential funder, or other would-be collaborator, please contact him here.
Misha Gurevich, $50K, to manufacture far-UVC lamps. Far-UVC is a type of ultraviolet light that kills germs rapidly; in a room with correctly-installed far-UVC lighting, viruses and bacteria die before they can reach another host, and the spread of contagious diseases plummets. In a world where this technology reached its full potential, respiratory pandemics like flu and coronavirus would cease to occur. Until now, these lamps have been limited to a few research prototypes. Last year, an ACXG-sponsored study worked to establish that they are safe for human use; results were reassuring. The next step is to produce them at scale as a consumer product for use in schools, daycares, and houses. Misha’s company Aerolamp has an early developer’s kit lamp on sale now, and is looking to hire an industrial designer experienced in safety and compliance who can help them transition to a mass-manufacturable version. If that’s you, get in touch with them here. Misha is a personal friend and a longtime ACXG evaluator; due to conflict of interest, this grant is being covered in conjunction with an outside funder.
Dan Elton, $25K, for a “metascience observatory”. Dan wants to use AI to “generate metrics that shed light on the health of science” - what percent of studies in different fields are retracted, challenged, successfully reproduced, etc. Although many people are monitoring reproducibility, Dan thinks he can develop an AI pipeline to do it at massive scale, eventually expanding to all of science.
Elaine Perlman, $94K, to continue lobbying for kidney donation incentives. Elaine works with Waitlist Zero and the Coalition To Modify NOTA to promote the End Kidney Deaths Act, which offers valuable tax credits to kidney donors. They estimate this bill could save 100,000 lives over the next decade, and save the government $50 billion/year (dialysis is very expensive, Medicare currently covers it, and transplantees would no longer need it). Since our previous grant last year, the EKDA has been cosponsored by 29 members of Congress, discussed in the Journal of the American Medical Association, and profiled in the LA Times. The prediction markets are down to only 25% chance it gets passed this year, but I’m optimistic about 2026 - 2027
Manoj Nathwani, $12K, for Allo Munganga, a telemedicine platform for the DRC. There is ongoing conflict in East Congo, and “all physicians have fled after armed groups took over”. But there are still some working pharmacies and labs, and Manoj wants to pull together telemedicine infrastructure so patients can continue getting diagnoses, lab tests, and prescriptions. He has partnered with a local medical group and will be using our money to buy technology, pay salaries and offer free consultations to patients over the phone.
Jacob Witten, $80K, to research mRNA for pulmonary disease. We are proud to fund Jacob’s effort, but his startup is still in stealth and we can’t provide further details.
Thomas Briggs, $5K, for the Center for Educational Progress. CEP was founded by Jack Despain Zhou, who you may know better by his blogging pseudonym TracingWoodgrains; he is currently on leave as he pursues his legal training, but will return next year. The Center advocates effective pedagogy, especially ability tracking, ie letting faster and slower students each move at their own pace. In practice, this seems to mean a lot of legal briefs telling San Francisco why they shouldn’t ban algebra in middle schools. We support their work and are happy to fill their suspiciously-low funding request.
Simon Chen, $25K, for automated forecasting work. There are already LLMs that are pretty good at forecasting; Simon wants to do the “unfun” work of optimizing them. In his proposal, different model parameters like prompt, temperature, AI model, etc, “form coalitions” based on past performance, with the exact details optimized against past successes. Our grant pays for his time and compute, and he hopes that once he has a working prototype he can get more money by winning forecasting tournaments.
Felix Nwose, $10K, for fish welfare in Nigeria. Felix is an aquaculture specialist who plans to hold workshops to train local fish farmers in techniques that improve conditions and lower mortality.
Jorge Bastos, $70K, for AI that curates bio datasets. There are exabytes (= 1 billion gigabytes) of high quality biology data; most of it goes unused because it’s not compatible with other datasets or tools. Jorge’s startup, Covalent, uses AI to put these in standard machine-readable format. At the very least, this would save biologists thousands of hours per year; in a best-case scenario, it could bring forward the golden age of AI-assisted biology predicted by people like Dario Amodei.
Greg Sadler, $65K, for Good Ancestors Australia. Our first grants round in 2021 supported ACX commenter Nathan Ashby beginning policy work in Australia. His work eventually evolved (it’s complicated) into GAA -now one of Australia’s most influential AI safety organizations, working with the public, MPs and their staffers to incorporate the x-risk/alignment perspective into Australian AI policy and legislation. We are excited to fund their continued operation. Australia is also a key base for building influence in tiny Pacific Island nations; although these may not have cutting-edge AI industries, they collectively form a powerful bloc in one-country-one-vote forums like the UN.
Yonatan Grad, $78K, for research and advocacy on antibiotic resistance. Recently, pharma has developed new antibiotics. Standard practice suggests that doctors hold these in reserve, deploying them only against bacteria that have develop resistance to all the old ones. Yonatan, a professor of immunology at Harvard, has models suggesting that the optimal strategy is more complicated, and might differ by disease: in some cases, you should hit the pathogen with everything you have all at once, to prevent resistance from developing in the first place. Our grant funds his work improving his models and building connections with medical policy-makers.
Matthew Loftus, $45K, for an HIV/TB clinic in Kenya. As a doctor working “on the ground” in developing world medical care, Matthew was a key voice in the recent campaign to save PEPFAR funding. This campaign ended in partial victory, with most key programs maintained but some infrastructure and support funding scaled back. Matthew will spend most of our grant integrating his local hospital’s HIV/TB clinic with their main operations (futureproofing them against infrastructure/support cuts), and the rest of it to continue his role as an influencer and educator about foreign aid and developing-world medicine.
Chetan Kharbanda, $30K, to help build an effective altruist ecosystem in India. Although some rich people like Bill Gates start with strong opinions on what they want to fund, much of the HNWI philanthropy space depends on people who go around to wealth management firms and help the rich understand their charitable options. Chetan and his cofounder want to make sure that India’s millionaires - 33,000 of whom get minted every year - are exposed to EA principles and opportunities. Their current project is an animal welfare funding circle; if you’re in India and interested in participating, please let them know.
Kurtis Lockhart, $85K, to continue research into African urbanism. Africa suffers from a sort of malignant anarcho-tyrannical NIMBYism, where the ability to build good urban infrastructure like roads, sanitation systems, or apartment buildings is gated behind an impossible series of permits and applications that the government never grants, but it’s easy to build endless illegal shantytowns. Kurtis runs the African Urban Lab, a joint project of the African School of Economics in Zanzibar and the Charter Cities Institute. They hope to build an “African YIMBY movement” within African academia/government to improve the situation in time to help the 900 million new people predicted to move to African cities in the next 25 years. Our grant will fund road planning advocacy, a satellite-based land tax system, and improvements to African universities’ urban planning curricula. Read more about Kurtis’ agenda in his Asterisk article, Yes In My Bamako Yard.
Bryan Davis, $50K, for software tools that speed FDA applications. Critics often focus on the expensive studies required for FDA approval, but those at least have a public interest benefit; a less-well-known hurdle is the logistics of the application itself, which use “an opaque, Adobe-only file in a deprecated format that resists integration into collaborative workflows”; most companies hire expensive consultants to explain the software to them rather than risk ruinous errors. Bryan and his team are working on open-source software that integrates with the FDA’s preferred format and automate the “application consultant” role. Our grant pays for their MVP.
Eli Elster, $13K, to research traditional psilocybin use in Africa. Psilocybin, aka magic mushrooms, is in the process of being integrated into mainstream psychiatric practice; it is already approved for treatment-resistant depression in Australia, and undergoing (currently promising) FDA trials in the United States. Much of what we know about the preparation and administration of psilocybin - including widespread ideas about “set and setting” and “integration” - comes from traditional use by the Mazaetec Indians. In 2023, anthropologists discovered that traditional healers in Lesotho, Africa also use psilocybin mushrooms - the first time such a practice has been found in the Old World - and that they seem to prepare and administer it differently from the Native Americans. Eli and his collaborator Betsy Sethathi conducted the first in-depth fieldwork on the topic earlier this year; our grant funds a return trip to Lesotho to further investigate their ethnobotanical practices and see if we can learn anything from them.
JD Bauman, $40K, to help fund Christians For Impact. Christians are a large and charitably-inclined demographic, but tend to bounce off the effective altruist movement after we start talking about becoming bodiless immortal machine-gods. JD and his team of Christian EAs network with churches and introduce them to everything else - all the ideas about how to realign one’s life around helping people in need. They have a blog, a career counseling network, and a conference that recently scored a guest appearance by the Archbishop of Canterbury. Our grant helps them publicize and expand their career counseling work.
Bengusu Ozcan, $30K, to raise awareness on AGI among EU policymakers. We were encouraged by the reception of the AI 2027 scenario in the United States. Bengusu’s team at the Center for Future Generations works on producing similar scenarios in Europe and explaining them to EU policy-makers. Our grant helps pay for their facilities, administrative overhead, and a quantitative dashboard add-on to the scenario presentations.
Sam Glover, $60K, to fight for free speech in the UK. These are dark times for UK speech on both sides of the aisle: the left is upset that speaking in support of Palestine Action is now considered an act of terrorism, and the right is upset about arrests for racist tweets. So far, pushback has been siloed by cause and partisan affiliation. Sam and his two co-founders are early-career bloggers and aspiring public intellectuals who want to build a united nonpartisan free speech movement. They’re still in stealth, but I’ll promote their website as soon as it becomes public.
Saeed Ahmad, $10K, to build an epidemic reporting system in Liberia. Liberia has been on the forefront of some recent pandemics, including Ebola and monkeypox. Saeed is currently based in Liberia, and wants to build infrastructure to translate local rumors about unusual diseases into reports to the national health authorities, including community reporters, phone hotlines, and social media.
Subhash Sadhu, $23K, for low-cost ultrasound scanners. Unlike fancier imaging modalities like CT, ultrasounds are safe and portable; there’s no reason to gate them behind hospital access, and broader ultrasound data could improve understanding of diseases from acid reflux to reproductive issues. Subhash and cofounder Siva Swaminathan are building a cheap wearable ultrasound “patch” potentially suitable for developing countries, people in inaccessible regions, researchers, or biohackers, plus an AI interpretation system. Our grant helps pay for components and a preliminary study to build a prototype.
Nuno Sempere, $50K, for disaster forecasting and response. Nuno runs Sentinel, a team of superforecasters which tracks incipient disasters (pandemics, wars, etc) and brainstorms pre-hoc and post-hoc responses. Their model for response are groups like VaccinateCA, a small team of Californians who noticed that the state’s COVID vaccine policy was disorganized and made a site that helped connect people with spare vaccination capacity. You can see their blog here. Nuno is an ACX Grants evaluator; due to conflict of interest, this grant is being covered in conjunction with an outside funder.
Alejandro Acelas, $24K, automated customer screening for DNA orders. When researchers or biotech companies need DNA, they send the sequence to a synthesis company, which then sends them back a finished product. But terrorists can also use these companies to make bioweapons on demand. Some (not all) companies check whether the sequence looks like a bioweapon first, and if so, spend hours manually trying to figure out if the customer has a legitimate reason to want such a thing. Alejandro and his cofounder are working on an AI screening tool to automate the latter part of this process.
G, $50K, for a secret project involving snakes. Of all factory-farmed animals, one of the worst lots goes to the hundreds of millions of mice raised each year as snake food. During life, they are confined in plastic bins with minimal enrichment and no ventilation; after reaching maturity, they are killed by crude methods like gassing or freezing, or transported directly to a a brutal death at the fangs of one of nature’s scariest predators. G is working on a techno-solution, but her effort is still in stealth and we can’t provide further details.
Harry Warne, $25K, for AI assisted speech amplification. Millions of people, including Harry, have vocal cord diseases that prevent them from speaking above a whisper. Microphones can make their voices louder, but not clearer - an amplified whisper sounds nothing like normal speech. But this type of problem is a good fit for AI, which can be trained to recognize dysphonic speech and match it to its normal equivalent. Harry has a prototype battery-powered voice converter which outputs normal-sounding speech almost fast enough to be useful. Our grant will help him clear the last few hurdles and bring it to market.
I’m still working on shopping a few more projects to VCs, and I haven’t gotten to the impact certificates yet. I’ll announce those once they happen.
Credits
A huge thanks to everyone who supported ACX Grants.
First and most important, our funders: Craig Falls, Calvin French-Owen, Shauna Kravec, Anton Makiievskyi, Geoff Price, Adam Winkel, and several people who asked to remain anonymous.
Second, the Manifund team. Manifund, a charitable spinoff of Manifold Markets, handled our funds, disbursement, infrastructure, and miscellaneous coding needs. Special thanks to Austin Chen for taking point on this.
Third, the many expert evaluators who volunteered their time to look over shortlisted grants, discuss them with the rest of the team, and help us settle on a final list. I still haven’t finished getting everyone’s permission to list their names, and will be expanding the “et al”s as these come in. By subject:
Generalist: Austin Chen, Misha Gurevich, et al
Biology: Metacelsus, Sarah Constantin, Ruth Hook
Health: Simon Grimm, Trevor Klee, Eryney Marrogi
Animal: Ozy Brennan, et al.
Forecasting: Austin Chen, Nuno Sempere
Development: Meir Brooks, Andrew Martin, et al
AI: Oli Habryka, et al
Metascience: Stuart Buck
Meta: Clara Collier, et al
External consultants: Paige Brocidiacono, Jay Lubow, Neel Nanda, John Schilling, Alex Turner, Robert Yaman, et al
Fourth, everyone who deserves credit but whom I failed to thank above, for various reasons. These include:
The 100 or so (!) people who offered to help as evaluators/consultants, but who we didn’t end up calling on because there weren’t any grants that were a clear match for their area of expertise.
The many people who offered to give special services like accounting and consulting to ACX grantees. I’ve gathered this into a directory and put it in the grantees Discord server. If you should have access but don’t, email me.
The people who offered funding after I stopped checking the funding offers form (sorry!) or who were considering offering funding but asked me technical questions about Manifund that I failed to follow up on appropriately.
Evaluators who didn’t answer my short-notice question about whether I had permission to list their names here.
The lawyers who worked with us or recommended colleagues to work with us, - in some cases pro bono, in others more-than-earning their fees.
VCs, representatives of other philanthropic foundations, and friendly professionals who I’m still gradually working on following up with.
Finally, thanks to all applicants. It’s a joy to see how many people are still coming up with big ideas, even if I can only fund a small fraction.
If any of you are unhappy with how you have been credited or not-credited, please email me at scott@slatestarcodex.com.
The next ACX Grants round will probably begin late 2026 or early 2027.