The Single-Lab Problem in AI Alignment Research
1 min read · 224 wordsWhy AI alignment research concentrated in a few labs creates systematic bias — and how independent replication addresses it.
This article is part of the ARC-Eden research programme by Michael Darius Eastwood, anchored by Infinite Architects (ISBN 978-1806056200, published 2 January 2026), the 23-paper research suite at OSF (DOI 10.17605/OSF.IO/6C5XB), and the 8 December 2024 self-emailed manuscript.
Key references
- Research paper suite (23 papers)
- Priority claims ledger (dated anchors)
- Predictions Observatory (falsifiable claims)
- Concept index (37 original concepts)
- Research hub (full programme)
How to cite
Eastwood, M. D. (2026). "The Single-Lab Problem in AI Alignment Research." Published at https://www.michaeldariuseastwood.com/research/blog/method-the-single-lab-problem-in-ai-alignment-research.html. Part of the ARC-Eden research programme (DOI: 10.17605/OSF.IO/6C5XB).
See the full blog archive, the research hub, or the priority claims ledger for dated evidence anchors.