AI alignment

Field within which this work has implications for evaluating alignment progress.

Neighborhood — ranked by edge-count

paper

framework

Self-Other Overlap (SOO) Fine-Tuning
about
The central framework proposed in this paper: aligning AI internal representations of self and others to reduce deceptive behavior

claim

Behavior under observation differs from behavior in deployment
associated_with
Epistemic principle: benchmarked safety cannot be assumed to hold in real-world use.

concept

Alignment
related_to
The goal of making model behavior match human values and intentions, often addressed during post-training.
AI Alignment and Safety
related_to
The broader domain for which ESR has dual implications: resistance to adversarial manipulation vs. interference with safety interventions
Ai Alignment Problem
related_to

cosine ≥ 0.65 · no typed edge

Entities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.

How Can We Ensure Alignment Between Artificial Intelligencequestion0.860
How Can We Align Powerful Ai Systems Whenquestion0.820
Alignment Problemconcept0.811
The problem of ensuring AI systems adopt values compatible with human welfare — argued to be a perennial problem already present in child-rearing
Alignment Functionconcept0.808
A learnable invertible transformation in DAS that maps neural representations to a basis aligned with causal variables
AI alignment with human interests is good for both humans and potentially for AI systems, providing a basis for overlapping consensus on alignment.claim0.808
Second mutual-benefit model example from Long (2025b).
Alignment Scoreconcept0.807
GPT-4o scored 0-100 metric where lower values indicate more misaligned behavior on open-ended evaluation prompts
Alignment Typeconcept0.795
The only statistically significant predictor of koan battery scores (p=0.006); includes Constitutional AI, RLHF, SFT, roleplay, empathy
Representational Alignmentconcept0.793
Measure of similarity between the similarity structures (kernels) induced by two different representations