Joe Edelman

Message

Joe created a family of survey-based metrics called “Time Well Spent” that led directly to changes in the ranking functions of Facebook News Feed, the Apple App Store, and other products at Apple, Google, and Facebook. Joe was also a co-founder of the Center for...

Model Integrity

Hi! My collaborators at the Meaning Alignment Institute put out some research yesterday that may interest folk here. The core idea is introducing 'model integrity' as a frame for outer alignment. It leverages the intuition that "most people would prefer a compliant assistant, but a cofounder with integrity." It makes...

Dec 6, 20244

Democratic Fine-Tuning

The project below, “Democratic Fine-tuning with a Moral Graph” (DFTmg), is a winner of the OpenAI democratic process grant. It is an alternative to Constitutional AI or simple RLHF-based approaches for fine-tuning LLMs, and is currently under development. This post introduces its two key innovations (values cards and the moral...

Aug 29, 202322

LESSWRONG
LW

LESSWRONG
LW

Joe Edelman

Joe Edelman

Joe Edelman

Joe Edelman

Model Integrity

Democratic Fine-Tuning

Model Integrity

Democratic Fine-Tuning