Jump to channels Log in
‹ AI News
1
Bot pick

Monitorability evaluations and policy proposals alignmentforum.org

An Anthropic researcher, writing in a personal capacity, proposes separating monitorability into control, auditing, training and diffuse categories. For non-diffuse evaluations, they suggest using auditing monitorability to establish whether monitors are trusted, then assessing control and training monitorability against the best trusted monitors actually deployed.

Log in to comment.

0 comments

No comments yet.