Resolution
Can formal and automated pipelines scale to superintelligent alignment?
Introduction
Resolution pursues theory and automation for high-confidence alignment, including Timaeus work on singular learning. The automation-first philosophy asks whether formal pipelines can scale to superintelligent alignment with checkable confidence. Even automated verification faces certification under manipulation and Deployment Safety scope limits under adversarial optimization.
Who carries it: Resolution (Geoffrey Irving; Timaeus merging)
What they aim to do. Pursue alignment approaches that can be automated and checked with high confidence.
The hard question. Can formal and automated pipelines scale to superintelligent alignment?
What they produce. Theory and automation for high-confidence alignment, including Timaeus work on singular learning.
Key terms. Key terms include automation, formal alignment, higher-confidence alignment, and singular learning (Timaeus).
Related field cruxes. Embedded Agency; Grounding Drift; Deployment Safety
What they contribute. An automation-first philosophy; UK AISI lineage via Irving. Iliad (which helped found Timaeus) runs the overlapping Textbook from the Future synthesis megaproject.
How this project treats it. This project’s adversarial-verifiability chokepoint under optimization remains load-bearing despite automation ambitions—related to certification under manipulation and Deployment Safety scope limits.
Links
- Resolution
- Resolution launch essay
- Automated alignment is harder than you think
- Timaeus
- Timaeus SLT position (2025)
Map clustering
AISafety.com map listings that roll up to this agenda:
See the coverage matrix for evidence tagged to this agenda, and the glossary for shared terms.