whitehatStoic
whitehatStoic
Values-based model vs. large language models
1×
0:00
Current time: 0:00 / Total time: -25:34
-25:34

Values-based model vs. large language models

There is something in this blogpost that may solve the alignment problem
Miguel de Guzman
Feb 05, 2023
Share

Read the full blog here:

whitehatStoic
On values-based model vs. large language models
Introduction To ask ChatGPT about how it separately perceives ethical models from how he was trained and see improvement in its own models, I believe the best approach is to break these large language models into components. This will give ChatGPT a unique and ambitious perspective on the evolution of its models and ethical situations it can come across …
Read more
2 years ago · Miguel de Guzman

Discussion about this podcast

whitehatStoic
whitehatStoic
Exploring evolutionary psychology and archetypes, and leveraging gathered insights to create a safety-centric reinforcement learning (RL) method for LLMs
Listen on
Substack App
Spotify
RSS Feed
Appears in episode
Miguel de Guzman
Recent Episodes
Token Doppleganger
 • Miguel de Guzman
Engineering a Life Preserver AI: Navigating the Complexities and Challenges
 • Miguel de Guzman
Engineering an AI as a Life Preserver—A Technical Deep Dive
 • Miguel de Guzman
Engineering the Collective Consciousness: Building an AI Unconscious to Safeguard Life Across the Cosmos
 • Miguel de Guzman
Exploring the Alignment of Superintelligent AI with Jung's Collective Unconscious
 • Miguel de Guzman
Corrigibility: A Hindrance to True AI-Human Cooperation?
 • Miguel de Guzman
The Brittle Balance: Navigating the Fragility of Human-AI Collaboration
 • Miguel de Guzman
Situational Awareness and Its Relevance to AI Alignment
 • Miguel de Guzman