Thank you for writing this up Stan! Hope this gets a bit more traction now - this could be a really good mechanism. A group of us are also thinking about AMC for AI Safety credts as well. If somebody is interested to collaborate on this, please get in touch.
By the way, the paper "Advance Market Commitments: Insights from Theory and Experience", which seems relevant to this post, was evaluated by The Unjournal – see https://doi.org/10.21428/d28e8e57.652a843a. Please let us know if you found our evaluation useful and how we can do better – [email protected]
[Linkpost] How to start an advance market commitment (AMC)
Advance market commitments, or AMCs, are promises to buy or subsidize something in the future, if someone can invent and produce it. The purpose of an AMC is to provide a financial incentive for innovators to develop and scale products that address important societal needs but lack a natural commercial market.
Past AMCs
Pneumococcal Vaccines In the 2000s, hundreds of thousands were dying of pneumococcal diseases every year. However, because the deaths were occurring in low-income countries unable to pay for new treatments, there was little investment in developing new vaccines.
In 2007, a group of governments and philanthropists pledged $1.5 billion through an AMC run by Gavi to accelerate pneumococcal vaccine development. The AMC subsidized vaccine doses beyond what poorer countries could afford, to incentivize pharmaceutical companies like Pfizer and GlaxoSmithKline. By 2011, these companies had developed qualifying vaccines and signed contracts for large-scale production. The AMC accelerated vaccine development and distribution by around 5 years, saving an estimated 700,000 lives.
COVID-19 Vaccines As part of Operation Warp Speed in 2020, the U.S. government issued $900 million in guaranteed purchase orders for COVID-19 vaccine doses from pharmaceutical companies. This provided an incentive for companies to invest in vaccine development despite uncertainty around which vaccines would succeed. The purchase guarantees ensured companies would still have a market even if other companies developed vaccines first.
Frontier: a carbon-removal AMC
Frontier is a $925 million advance market commitment launched in 2022 to accelerate the development of affordable technologies for removing carbon dioxide from the atmosphere at scale. It aims to bridge the gap until a long-term market for carbon removal emerges, by providing funding commitments for future carbon removal. Frontier aims to catalyze the carbon removal industry by sending a strong demand signal to motivate entrepreneurs, investors, and researchers to prioritize carbon removal innovation, while also advocating for policies to create a durable, global market for permanent carbon removal after 2030.
Before 2022, only $30million had ever been spent on carbon removal. Frontier's $925million represents a new era for carbon removal tech, but billions more in investment will probably be required before the unit cost of carbon removal becomes low enough for a mass market to emerge.
Where else could AMCs be used?
AMCs have been proven as a way to accelerate vaccine development, but they may also be a powerful way to advance other critical technologies:
Headline finding: I audited 17 AI Safety Talent programmes. Zero of 17 have published any comparison group, rejected-applicant follow-up, matched control or randomisation. Not one. Every programme that mentions a counterfactual does it by asking participants to self-report.
Background
At least $70 million...
In celebration of still being alive and fighting, we are giving away 1,000 Amazon e-books of “If Anyone Builds It, Everyone Dies”. Feel free to send a copy to yourself, a loved one, or a friend—we need all hands on deck.
Today marks exactly one year since If Anyone Builds It, Everyone Dies: Why Superhuman AI Would Kill Us All, by Eliezer Yudkowsky and Nate Soare...
A year ago I posted that YCombinator (YC) companies didn’t seem to be growing faster since the release of ChatGPT in 2022. I reran that experiment and found that 2023+ YC companies are arguably growing a bit faster than pre-2023 companies (including AfterQuery, ...
Thank you for writing this up Stan! Hope this gets a bit more traction now - this could be a really good mechanism. A group of us are also thinking about AMC for AI Safety credts as well. If somebody is interested to collaborate on this, please get in touch.