Everything you care about in one place
Follow feeds: blogs, news, RSS and more. An effortless way to read and digest content of your choice.
Get Feederalignmentforum.org
AI Alignment Forum
Get the latest updates from AI Alignment Forum directly as they happen.
Follow now 23 followers
Latest posts
Last updated 1 day ago
Announcing the ARC White-Box Estimation Challenge
4 days ago
ARC has teamed up with AIcrowd to launch the ARC White-Box Estimation...
Testing Gemini models for scheming tendencies
8 days ago
As AI models become increasingly capable and autonomous, keeping them safely aligned...
Advice for making robust-to-training model organisms
9 days ago
We’d like to develop training techniques that work when applied to future...
Eval Cooperativeness May Be a Scalable Mitigation for Eval Gaming
10 days ago
Behavioral evaluations may become worthless, which we think would be a disaster...
Full automation of AI R&D probably yields a large speed up even without a software-only singularity
10 days ago
This is a somewhat technical note By "software-only singularity", I mean that,...
Looking for backdoors in Jane Street LLMs
15 days ago
I am going to talk about my experience in the Jane Street...
The Case for Evaluating Model Behaviors
17 days ago
Most evaluations of AI systems focus on their capabilities: how good they...
Risk reports need to address deployment-time spread of misalignment
22 days ago
Risk reports commonly use pre-deployment alignment assessments to measure misalignment risk from...
Mechanistic estimation for expectations of random products
22 days ago
We have developed some relatively general methods for mechanistic estimation competitive with...
The safe-to-dangerous shift is a fundamental problem for eval realism; but also for measuring awareness
23 days ago
1) The safe-to-dangerous shift is a fundamental problem for eval realismSuppose we...
Empowerment, corrigibility, etc. are simple abstractions (of a messed-up ontology)
26 days ago
1.1 Tl;drAlignment is often conceptualized as AIs helping humans achieve their goals...