Program Summary
Rapid advancements in large pre-trained models capable of generating scientific content pose a growing challenge to information integrity and authenticity. A fast, scalable, compelling generator for inaccurate scientific content could disrupt the U.S. technology base in times of crisis, such as pandemics, disaster relief, and military events, or upend the global race for technological dominance in critical areas.
Furthermore, false capability claims can also have significant implications for national security and international relations, as they can obscure the true nature of military actions and capabilities. However, determining the feasibility of advanced technologies, such as quantum computing, often requires deep technical expertise, access to detailed information about the technology, and the ability to conduct testing or independent verification. Overcoming these challenges requires developing new, robust methods to automatically review, reason, verify, and evaluate capability claims, especially in sensitive areas surrounding national security and defense.
The Scientific Feasibility (SciFy) program aims to develop computational methods for measuring the feasibility of claims, thereby enabling accurate assessments of scientific content. The program will focus on claims that express scientific and technological capabilities and demonstrate that the scientific feasibility of claims can be determined by using automated reasoning to decompose claims into constituent, verifiable parts --thereby paving the way for a new era of scientific assessment.
SciFy seeks to produce methods that perform well beyond current automated fact-checkers, recognizing that feasibility assessment is a complex process that requires breaking claims down into constituent components. Automatically assessing each component may involve identifying and using existing technological advancements, foundational scientific principles, data, software, simulation results, and current industry standards or benchmarks. This assessment necessitates the development of sophisticated automated techniques capable of managing the rapid expansion of evidence, ensuring that the synthesis and explanation of this evidence is both efficient and reliable. It is also necessary to determine whether the claimed technological capability, while theoretically possible in parts, is practical and realistic when considered as a whole, which may require evaluating logical consistency, system integration, and compatibility considerations.
SciFy is a 32-month, one-phase program consisting of three technical sprints in the material science, AI, and quantum computing domains. These domains are intended to represent a progression in scientific complexity.
A Broad Agency Announcement solicitation with all program details is available on SAM.gov at this link: https://sam.gov/opp/c4c8d4554adc42a888f22d3c2047a3ae/view.