GPT-4 is now able to write its own code and execute python scripts

detroit become human incoming

https://techfundingnews.com/openai-backs-1x-technologies-in-23-5m-funding-to-commercialise-humanoid-robotics/

imagine the bill for OpenAI API usage

It seems like StabilityAI has found a new founder. Let's hope they won't turn into another ClosedAI.

ps.sorry for not posting for a long time, my job has been keeping me very busy lately

Zip-NeRF: Anti-Aliased Grid-Based Neural Radiance Fields

this is not a recorded video - it is fully rendered from a neural model


arxiv
page

Stability AI announces Stable Diffusion XL beta for API and DreamStudio

it's not required for you to use depth maps to fix hands in ControlNet anymore, it can be done with pose2img in ControlNet 1.1

Harmony-Rhythm Disentanglement audio remixer plugin
Sadly, only for macOS

“Generative Agents: Interactive Simulacra of Human Behavior”
The paper aims to introduce a new approach to creating believable proxies of human behavior that can be used in various interactive applications such as immersive environments and prototyping tools . The authors propose a generative agent that can imitate human actions and interactions in a virtual environment . These agents can form opinions, socialize, and plan activities , making them useful for various applications such as immersive environments and prototyping tools . The evaluation shows that the generative agents produce believable individual and emergent social behaviors, such as forming relationships and coordinating group activities.

arxiv
project page

BabyAGI as a plugin for chatGPT.

BabyAGI is an open-source platform that draws inspiration from the cognitive development of human infants to facilitate research in various fields, including reinforcement learning, language learning, and cognitive development. It is a simplified AI-powered task management system that leverages OpenAI and Pinecone APIs to create, prioritize, and execute tasks.

A follower provided some valuable reading material.

Jascha Sohl-Dickstein has written a blog post titled “The hot mess theory of AI misalignment: More intelligent agents behave less coherently” . In this post, Jascha Sohl-Dickstein argues that misalignment risk is not about expecting a system to “inflexibly” or “monomanically” pursuing a simple objective but about expecting systems to pursue objectives at all. The objectives don’t need to be simple or easy to understand.

The AI texture/sprite generator package for Unity

OpenAI released a implementation of Consistency Models

Stability AI released their language model