12,445 papers · continuously updated · last export: 10 Aug 2026livingmeta.ai
← Browse all papers
AI evidence extraction

AI for Science Needs Scientific Alignment

Savannah Thais, Roberto Trotta, Nathan Suri, Emily Sullivan, Viyan Poonamallee, Tanaporan Na Narong et al. · PhilSci-Archive (University of Pittsburgh) · 2026

AI-generated evidence extraction, verified across multiple analytical personas. Not a substitute for the peer-reviewed original.

9/10
Relevance
0/4
Quality (LMQS)
I
Evidence
0
Citations

Methodology & findings

Study design

Position paper with conceptual analysis and argumentative framework; hermeneutic and critical analysis of current AI-science integration practices

Main result

The paper identifies that "conflicting claims about fundamental capabilities, documented contraction of research toward AI-amenable problems, and benchmark-driven development disconnected from scientific needs" are problematic patterns emerging as AI investment in science grows. The authors argue that "science is an inherently valuable epistemic system oriented toward human understanding—not merely prediction—and that its value and reliability depend on social infrastructure that is now threatened by misaligned AI integration."

Research paradigm

Critical epistemology; philosophy of science; sociology of scientific knowledge

Author conclusions

The authors conclude that "the goal is not to constrain AI, but to ensure it serves the genuine aims of scientific inquiry." They propose establishing a new field of "scientific alignment" to address the gap, which requires both "technical alignment" ensuring "AI systems optimize for epistemic norms like traceability, self-consistency, and support for human comprehension" and "systemic alignment" for developing "governance structures that sustain science's social infrastructure."

Open questions raised

  • The authors identify gaps in current AI alignment and responsible AI frameworks, noting that neither "general AI alignment nor responsible AI frameworks adequately address" the alignment of AI to science's epistemic goals and values. They outline concrete research directions for scientific alignment as a new field of study.
  • The authors identify that "neither general AI alignment nor responsible AI frameworks adequately address" the challenge of aligning AI to science's epistemic goals and values. They outline the need for a new field of study (scientific alignment) and propose concrete research directions in technical and systemic alignment.
  • The paper identifies that realizing AI's potential for science while protecting science as a knowledge-producing institution requires alignment to science's epistemic goals and values—a challenge inadequately addressed by current frameworks. The authors propose scientific alignment as a new field of study with concrete research directions needed.
Data: not_statedCode: not_statedExtracted from: pdfAgreement 81%

Explore related topics

Responsible AI UseAI GovernancePosition Paper

Related papers

← Browse all papers