Toward an Engineering of Science: Rebalancing Generation and Verification in the Age of AI

2026-05-11

AI Summary

This work addresses the growing risk of epistemic pollution in scientific research stemming from AI-generated content, which undermines implicit filtering mechanisms historically afforded by high production costs while leaving verification costs unchanged. Framing knowledge production as an engineering problem, the paper proposes replacing traditional narrative papers with structured “blueprints” that explicitly disentangle claims, evidence, assumptions, and definitions using typed graph models. This approach shifts verification upstream by creating machine-readable, verifiable intermediate representations of scientific work. Although it incurs modest additional overhead during initial content generation, it substantially reduces the complexity of subsequent validation. The authors present a proof-of-concept prototype demonstrating a viable pathway toward reimagining scientific infrastructure for the AI era.

Abstract

AI systems can now cheaply generate plausible scientific artifacts such as papers, reviews, and surveys. This creates a risk of \emph{epistemic pollution} in our scientific systems, where unreliable but plausible-looking artifacts can accumulate faster than the system can filter them out. The problem is structural: the epistemic infrastructure of science was calibrated to a world where producing a plausible artifact required substantial expertise, labor, and time, so generation cost itself served as a rough filter; AI weakens that filter without comparably lowering verification cost. We argue that \textbf{AI-era science should treat this as an engineering problem: redesigning epistemic infrastructure to rebalance the costs of generation and verification}. The current paper-centered system makes verification expensive: papers compress long-context scientific logic into prose, forcing reviewers, human or AI, to reconstruct underlying argument structure before they can evaluate it. As one step in this direction, we propose \textbf{blueprints} as preliminary epistemic infrastructure: structured, decomposed research artifacts that represent claims, evidence, assumptions, and definitions as typed graph components. Blueprints are designed to trade an upfront generation cost for cheaper, more local, more distributed verification downstream. We have instantiated the proposal in a proof-of-concept prototype.

Problem

• epistemic pollution
• AI-generated science
• verification cost
• scientific infrastructure
• generation-verification imbalance

Innovation

• blueprints
• epistemic infrastructure
• AI-era science
• structured scientific artifacts
• verification cost