Preparing for Launch (ifp.org)

This foreword to The Launch Sequence argues that AI progress is already compounding exponentially across coding, math, tool use and scientific analysis, and that the United States—because of its dominance in the AI supply chain—must proactively shape how that progress unfolds. The authors stress path dependence: the order in which capabilities are developed (e.g., offensive vs. defensive cyber or bio tools) can materially change risks and outcomes. They frame shaping as both an opportunity to accelerate AI-driven advances like drug discovery and materials science and a responsibility to prevent misuse, noting that private incentives and current funding (roughly 2% of AI papers and ~$100M/year) leave safety and defense under-resourced. Technically, the essay offers four guiding principles—exploit the “jagged frontier” of capabilities, account for the costs of stalled progress, redesign scientific institutions, and adapt to (and reduce) deep uncertainty—and calls for concrete public investments, export-control-aware policies, and new R&D institutions to tilt development toward beneficial and defensive applications. Key implications: targeted interventions (changing data availability or bottlenecks) can flip whether defense or offense leads; structural barriers—locked datasets, slow regulatory approval, and misaligned funding incentives—must be fixed to realize AI’s public-good potential; and export controls alone won’t suffice, so the U.S. must act as the “R&D lab of the world” to both accelerate science and build asymmetric defenses.