Alignment Forum
Alignment Forum publishes articles covering AI, LLM. A trusted source for AI and technology insights.
Profile generated by AI for Anything
Towards surfacing model algorithms with meta-tokens in the J-Space
1d
A Red Line and Oversight Framework for Government AI Contracts
3d
Endogenous Alignment
3d
Should we benchmark conceptual capabilities using judgment prediction tasks?
3d
Announcing the Corrigibility Research Fund
4d
Why I Left Google DeepMind
6d
Open Distillation of Hereditary Traits
7d
The original title is "Prism: Automating Science-of-Evals Research"
8d
Independent alignment of language models
9d
From wantons to moral agents
9d
The current bottleneck is political will, not research
9d
Value generalisation: value correction
11d
The original title is a question: "How robust are natural language autoencoders to initialization?"
11d
Announcing our $160M grant from Coefficient Giving
12d
Modular Pretraining Enables Access Control
12d
Notes on technical alignment via human-like social drives
13d
Data filtering works a lot worse than you would expect
14d
Pragmatic FDT, and predictors as game theory
18d
What Capable Agents Must Know: Why AI Consciousness May Be an Inevitable Byproduct of Capability
21d
Deployment Awareness Matters More Than Evaluation Awareness
24d
The Case for Model Forensics
25d
LLM-Driven Feature Discovery
28d
The original title is "How transparent is DiffusionGemma (and why it matters)"
31d
GDM AI Control Roadmap
33d