Accelerating Returns Don't Solve Core Scientific Discovery Problems.

The concept of accelerating returns, popularized by Ray Kurzweil, posits that technological advancements, particularly in areas like compute and AI, interact to create self-amplifying, exponential progress. While this thesis effectively describes the rapid growth of executional and infrastructural capabilities, this paper argues it falls short of addressing the fundamental challenge of scientific discovery. Genuine scientific breakthroughs often depend on qualitative reasoning—the ability to discern when existing frameworks are inadequate and to conceptualize the next necessary intellectual leap. Current AI systems, as evidenced by their performance on benchmarks like ARC-AGI-3 where humans excel and AI struggles, still lack this flexible, qualitative reasoning capacity. This highlights a significant gap between AI's current abilities and human understanding. The paper introduces the Qualitative Engine for Science (QES) as a framework designed to address this missing capacity. It suggests that while accelerating returns explain quantitative acceleration, QES focuses on preserving, organizing, and making accessible the human wisdom inherent in the processes of scientific discovery. The value of QES is independent of the timeline for Artificial General Intelligence, emphasizing the enduring importance of human qualitative understanding.