Evals as triggers and why they failed

The Don’t Worry About the Vase Podcast is a listener-supported podcast. To receive new posts and support the cost of creation, consider becoming a free or paid subscriber.

* 00:00 - Introduction

* 01:52 - Promises, Promises

* 03:19 - Anthropic Responsible Scaling Policy v3

* 03:41 - That Could Have Gone Better

* 04:49 - I’m Just Not Ready To Make a Commitment

* 08:50 - So Cold, So Alone

* 13:07 - I’m Sorry I Gave You That Impression

* 20:59 - Fool Me Twice

* 24:41 - In My Defense I Was Left Unsupervised

* 27:27 - Drake Thomas Finds The Missing Mood

* 30:12 - Things That Could Have Been Brought To My Attention Yesterday (1)

* 31:48 - Things That Could Have Been Brought To My Attention Yesterday (2)

* 37:54 - What We Have Here Is A Failure To Communicate

* 41:10 - You Should See The Other Guy

* 44:04 - I Was Only Kidding

* 45:01 - They Can’t Keep Getting Away With This

* 45:39 - Damn Your Sudden But Inevitable Betrayal

https://open.substack.com/pub/thezvi/p/anthropic-responsible-scaling-policy?utm_campaign=post-expanded-share&utm_medium=web