This website requires javascript to properly function. Consider activating javascript to get access to all site functionality.
AI ALIGNMENT FORUM
Hypothesis Subspace
AF
Login
Hypothesis Subspace
A living collection of alignment proposals I'm exploring at
Refine
, a program hosted by
Conjecture
.
14
Oversight Leagues: The Training Game as a Feature
Paul Bricman
2y
0
7
Ideological Inference Engines: Making Deontology Differentiable*
Paul Bricman
2y
0
17
Representational Tethers: Tying AI Latents To Human Ones
Paul Bricman
2y
0
11
Interlude: But Who Optimizes The Optimizer?
Paul Bricman
2y
0
15
(Structural) Stability of Coupled Optimizers
Paul Bricman
2y
0
11
Cataloguing Priors in Theory and Practice
Paul Bricman
2y
0