AGI Alignment Experiments: Foundation vs INSTRUCT, various Agent

Here’s the companion video: Here’s the GitHub repo with data and code: Here’s the writeup: Recursive Self Referential Reasoning This experiment is meant to demonstrate the concept of “recursive, self-referential reasoning” whereby a Large Language Model (LLM) is given an “agent model” (a natural language defined identity) and its thought process is evaluated in a long-term simulation environment. Here is an example of an agent model. This one tests the Core Objective Function

Facebook AI's Multitask & Multimodal Unified Transformer: A Step Toward General-Purpose Intelligent Agents

Auto-GPT: Unleashing the power of autonomous AI agents

RLHF: Reinforcement Learning from Human Feedback, by Ms Aerin

Future Internet, Free Full-Text

Examining User-Friendly and Open-Sourced Large GPT Models: A Survey on Language, Multimodal, and Scientific GPT Models – arXiv Vanity

WizardLM (@WizardLM_AI) / X

AI Agency Challenge

My understanding of) What Everyone in Technical Alignment is Doing and Why — AI Alignment Forum

ChatGPT, Next Level: Meet 10 Autonomous AI Agents: Auto-GPT, BabyAGI, AgentGPT, Microsoft Jarvis, ChaosGPT & friends, by Maximilian Vogel, The Generator

Artificial Intelligence & Deep Learning, Fastformer: Additive Attention Can Be All You Need

AI is centralizing by default; let's not make it worse — EA Forum

Exploring AGI: The Future of Intelligent Machines

AI Safety 101 : AGI — LessWrong

Why AI alignment could be hard with modern deep learning — EA Forum

OpenAI Launches Superalignment Taskforce