Statement from Economists on the Importance of Open Source AI

We, the undersigned economists and associated experts, believe that open source AI could be a critical driver of innovation, economic growth, and shared prosperity around the world if companies and governments promote its adoption rather than restrict its availability. Although there are many ways to define open source AI, when AI systems promote reuse and interoperability, lower the barrier to entry for startups, and increase corporate transparency and accountability, they have the potential to accelerate innovation and competition in the global economy.

History serves as a guide. As the commercial internet was coming to fruition, many politicians were concerned that widespread adoption of encryption would enable malicious actors to share information with each other. Eventually, the promise of international e-commerce, which would depend on strong, open-source encryption, made clear the economic benefits of an open and secure Internet. Companies and governments supported the fledgling internet that has created today’s booming digital economy. While there continue to be concerns over the widespread availability of encryption tools, most observers agree that the risks posited years ago have not materialized in a way that compares to the economic benefits the open internet has brought the world.

Today, it is reasonable to expect a similar positive impact from open source AI. We believe open source AI, appropriately defined, will generate many global economic benefits. For instance, it will lower the barrier to entry of innovative AI firms to enter with entirely new products. New entrants may further challenge the dominance of market incumbents, and thereby increase competition in the AI space. A wider range of organizations will be able to participate in building custom AI applications to improve their operations and better serve customers. For example nonprofits and social sector organizations may find the cost of closed AI models to be prohibitive, and the decision to procure access to expensive AI models may be difficult when the intended application is new and unproven. The transparency and collaboration that is fostered through open source AI will lead to new and better organizational structures to harness the benefits of innovation across sectors. Open source AI initiatives can utilize shared infrastructure and technological interoperability, which will increase global contributions, reach, and impact of these advancements. These are a few examples of the many effects of open source AI that, when taken together, have the potential to generate global economic benefits.

These myriad benefits are not without risks. Some companies and regulators have raised safety concerns related to open source AI. It is worth disentangling the two different types of safety that are broadly discussed: (1) Ensuring that developers and users of AI have the products and tools they need to make its application safe, and (2) Ensuring AI cannot be misused by malicious actors. For the former, open source AI is generally a boon, as it enables and incentivizes companies to compete to create safer and better products, and helps developers, researchers, regulators, civil society, and the public with auditing and assessing their safety. For the latter, however, there is vigorous debate about whether and how increased access to AI systems creates new risks.

Open source AI clearly has significant competition and innovation benefits. We acknowledge that different factions with the power to incentivize or block development of open source AI systems will need to manage these concerns to harness these benefits. Unique approaches to managing trade-offs between economic potential and these concerns will be vital for different sectoral regulators. For instance, competition and consumer protection agencies may be strongly in favor of open source AI, while national security entities may work to mitigate malicious usage risks in a way that has the least impact on the open source AI ecosystem.

Overall, for very low cost, we believe open source AI will have an outsized positive impact on the global economy. We urge governments and companies to proactively promote, rather than restrict, developments that foster open source AI.

We thank the Mozilla Foundation and associates for their support in convening the signatories of this letter.