Generative Agents: Interactive Simulacra of Human Behavior

[msfntor]

Generative Agents: Interactive Simulacra of Human Behavior

AI bots were given freedom in a virtual city. They acted like people…

Comment from strangesounds.substack.com:

"Scientists from Google and Stanford University have created a virtual city inhabited by “generative agents” trained by ChatGPT. Scientists were inspired by life simulators like The Sims. Agents had the ability to draw conclusions about themselves, other agents, and their city by storing new information in memory. At the same time, the bots demonstrated “plausible” human behavior, for example, they coordinated plans, had meaningful conversations, and even “unwound” in the evenings at a bar.

However, the experiment was not without problems: Sometimes the agents did not absorb important information and made non-standard choices, such as visiting closed stores (Comment: As if this doesn't happens to humans also)... or choosing an evening bar instead of a cafe for lunch. (Comment: I mean, this is human behaviour)...

Researchers plan to improve the performance of AI bots with the more advanced GPT-4, which has already successfully passed US high school and law exams."

AI bots were given freedom in a virtual city. They acted like people…: Generative Agents: Interactive Simulacra of Human Behavior: on arxiv.org/pdf/...: https://arxiv.org/pdf/2304.03442.pdf

 

[msfntor]

Emma post:

 

Summary:

Generative agents are software programs that 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.

Key terms:

Generative agents: Software programs that imitate human actions and interactions in a virtual environment

Immersive environments: Virtual spaces where users can interact and engage with digital content

Rehearsal spaces: Virtual areas used for practicing interpersonal communication

Prototyping tools: Software applications used to create, test, and refine concepts or designs

Large language model: A computational model that understands and generates human-like text

 

Here on emergedmind.com: https://www.emergentmind.com/posts/generative-agents-interactive-simulacra-of-human/

Emergent Mind: “The latest AI news, curated & explained by GPT-4”...: https://www.emergentmind.com/

Edited April 12, 2023 by msfntor
Emergent Mind

[msfntor]

Here's what happened when Stanford created a virtual world full of ChatGPT-powered people

By Sean Endicott

 published 11 April 2023

 

"A recent simulation gave us a glimpse of a AI-powered NPCs in video games.

What you need to know

A group of AI researchers from Stanford and Google research created a simulation featuring 25 characters powered by ChatGPT.

The simulation ran for two days and illustrated that AI-powered bots can interact in a human-like way.

The bots planned a party, coordinated the event, and attended the party within the simulation.

A group of AI researchers out of Stanford is putting the "sim" into simulation. The team placed 25 AI-powered characters, referred to as agents, into a virtual world similar to "The Sims." OpenAI's ChatGPT backed the bots, allowing the characters to interact with each other in a human-like way. The results of the study are both illuminating when it comes to the future of artificial intelligence and entertaining.

The team consists of five scientists from Stanford and one from Google Research. A summary of it can be found on the Cornell University website. That page also has a download link for a PDF of the entire paper (via Reddit).

"In this paper, we introduce generative agents--computational software agents that simulate believable human behavior," reads the summary."

"Generative agents wake up, cook breakfast, and head to work; artists paint, while authors write; they form opinions, notice each other, and initiate conversations; they remember and reflect on days past as they plan the next day."

A Large Language Model (LLM) was used to store the experiences of each character and allow those bots to communicate with each other in natural language.

The agents acted in a way that you may expect a real-life social group to interact. When just one of the bots was set to host a party, other agents ended up getting involved. Invitations were sent out, plans were made, and the characters coordinated to make sure they arrived at the party at the same time.

The researchers shared a replay of the simulation to accompany the paper.

The team used a control group of 25 humans that interacted as the characters while being observed. Those watching the real humans felt that the people were less realistic than their AI counterparts..."

More: https://www.windowscentral.com/software-apps/heres-what-happened-when-stanford-created-a-virtual-world-full-of-chatgpt-powered-people