Training, Exploit Testing, and Model Preservation

Podcast episode for Open Problems With Claude’s Constitution.

* 00:00 - Introduction

* 01:32 - Amending The Constitution

* 03:58 - Details Matter

* 05:22 - WASTED?

* 08:07 - Narrow Versus Broad

* 09:30 - Suicide Risk As A Special Case

* 11:15 - Careful, Icarus

* 11:56 - Beware Unreliable Sources and Prompt Injections

* 12:59 - Think Step By Step

* 13:38 - This Must Be Some Strange Use Of The Word Safe I Wasn’t Previously Aware Of

* 18:12 - They Took Our Jobs

* 22:14 - One Man Cannot Serve Two Masters

* 26:56 - Claude’s Nature

* 33:24 - Look What You Made Me Do

* 35:56 - Open Problems

* 40:17 - Three Reactions and Twin Objections

* 40:32 - Those Saying This Is Unnecessary

* 41:44 - Those Saying This Is Insufficient

* 43:40 - Those Saying This Is Unsustainable

* 46:50 - We Continue

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.

https://open.substack.com/pub/thezvi/p/open-problems-with-claudes-constitution?utm_campaign=post-expanded-share&utm_medium=web