ChatWorld: Co-pilot for Groups

• ChatWorld: Co-pilot for Groups
  • Quick Rundown on Upstreet
    • Layer 1: ChatWorld
  • Co-pilots for Working Groups
    • Benefits
  • Membership
  • Footnotes

---

Quick Rundown on Upstreet

Since 2021 (pre-ChatGPT) we have been leveraging AI in the writer's room to help us cultivate the vision of the Internet Cinematic Universe. It’s been a year since the Webaverse mint (2022). Since then there have been a few cool releases like MoeMate and Character Studio, an open source avatar customizer with AI integration and ability to export optimized VRM avatars. However, the big vision that helps tie everything together is Upstreet.

• Virtually embodied real-time immersive simulation
• Multi-agent Hyperspace Arena
• Infinite canvas you can play
• The Street
• Game world for humans and AIs
• Game you can watch
• Game that learns
• Infinite generative simulations / TV shows

These are all ways one can describe Upstreet, and the underlying tech powering it. Long story short; We have been working tirelessly to deliver the vision people have always wanted; a cross-universe world inhabited by humans and AIs, built upon foundations and principles that align with an open metaverse.

FAQ

• How is this related to Webaverse?
  • It's not, it's a community project with the founding team to ship the Upstreet AI metaverse.
• Who owns Upstreet?
  • The dev team and the community own the project together.
• How do I get land?
  • There was a redemption round to trade Webaverse NFTs for Upstreet title deeds. That window has since closed. If interested in building, hit up the discord - we have grants.
• Is Upstreet live?
  • The demo is playable now, including multiplayer: https://upstreet.ai/.

Upstreet and Webaverse are on different paths, but are heavily aligned. UpstreetAI now is the largest holder of Webaverse Genesis Passes.

Layer 1: ChatWorld

M3 and Autonomous Research Group recently supported a 28 page research paper called ChatWorld: Embodied Multi-Agent Simulation For Extensible Synthesis of Hypermedia, which goes into technical details about the systems architecture that currently powers Upstreet.

It covers:

• Multi-agent Simulation
• Human-AI Interaction
• Depth-aware / Diffusion based Virtual World Generation
• 2D Character Generation
• Audio / SFX generation
• Applications like AI companions, Interactive Game Worlds, Infinite TV Shows, Self-Programming agent, Multi-actor Decision Agent, and more

We see untapped potential in the creation of an infinitely generated virtual world, inhabited by AI-driven agents. Such an environment could function indefinitely even without a player base, given that the agents could be programmed to generate content and evolve the game autonomously.

---

Co-pilots for Working Groups

Leveraging AI with tools like ChatGPT still feels largely like a single player experience. Meanwhile, we've already discovered enormous value by making user interfaces multiplayer through UX like Discord bots.

M3 members have been actively collaborating across many metaverse interest groups focused on interoperable open standards. One of the growing challenges we have is how spread out things have become as this space grew, increasing overall coordination costs.

The vision

Lets supercharge our working groups with AI agents that assist us in onboarding, feedback, outreach strategy, and governance decision making within. Together we could form a dream team, or few depending on the department. The plan is to hook up AI pals with info on what we're all about – our goals, ongoing projects, and the treasure trove of past research.

Train LLMs on our research, including the ChatWorld paper itself, and hook into interfaces like avatars and discord bots. We can split different topics into different agent personas, each offering a unique perspective on the ideas one might be interested in. I've been gathering mountains of data from various metaverse interest groups like M3, OMI, MSF, and various DAO knowledge bases into a private GitHub repo for agent training, so each organization could have their own unique agents.

Sample ideas for agent personas

• Developer who has to implement it
• User who has to use it
  • We can have different profiles of such
• DAO who has to approve and market it
• Biz dev person
• Neal Stephenson who it must spark joy with

Additionally, we can boost our marketing by inviting these agents into our Discord calls and Upstreet hangouts, as well as make content together via panels and interviews. These AI agents can sit with us as part of our "Jedi Council", offering their two cents in our decision-making process. Or, we could go all "Shark Tank" with our AI crew to help evaluate grant proposals. Any of these options can also be streamed and/or recorded, giving our marketing efforts a much needed boost.

Btw did I mention we have a working prototype?: https://github.com/vinny-888/upstreet-ai-agents

Even if a proposal doesn't make the cut, applicants won't walk away empty-handed. They'll score some excellent feedback that'll level up their writing skills and maybe even spark some ideas that align with our goals and resources. It'll be like having various AI personal trainers, all with context by training on our data, help us get better at collaborating!

Here's another idea for down the line: we can give the AI proposal reviewers a teeny bit of voting power to sweeten the service. That way, folks will be eager to tap into their wisdom because, let's face it, humans usually take the path of least resistance for maximum gain. If humans are the only ones voting people might not bother asking for the council's advice.

I think the sooner we execute this the better. By being early adopters and refining various ChatWorld paper examples, it'll be easier for us to pitch others on how this tech will benefit their companies and organizations. It can also lead to enhance collaboration and alignment discovery with other orgs since we've gathered data from a number of public sources. Imagine M3 bots talking to OMI and MSF bots on Upstreet or ideating together in our discords, coming back to us with ideas and proposals.

Benefits

1. Enhanced Collaboration: By leveraging AI to help parse through vast amounts of research and content, we all benefit from improved decision-making, better information dissemination, and ongoing self-improvement, aligning with the Kaizen philosophy

2. Commemorative Tokens: Contributors receive commemorative tokens, incentivizing their participation and support. These tokens can be used to engage with Upstreet AI agents or potentially as a means of participating in governance decisions for things like agent personalities

3. Boost our Marketing Outreach AI agents casted into our virtual production pipelines can become useful in helping us produce marketing material as writers, concept artists, co-hosts, and virtual actors.

Membership

The NFT for the membership is a poster of the research paper itself that you can display at full resolution in your virtual home space. It serves as a mindmap for keeping the group on the same page.

If the party does not reach the ETH goal, the ETH will be returned back to holders. We first want to set aside a portion of funds to cover hard costs of running AI systems ($2k a month). Everyone that contributed gets a commemorative token, for which Upstreet AI agents may react differently to those holding it in their connected wallet.

Idea: Members get access to Dework space to create new suggestions and apply for bountys

---

Footnotes

1. What inspired early explorers to take the dangerous journey across the great oceans? We need more than a technological foundation to connect virtual worlds together. The open metaverse is also a cultural phenomenon emerging in terms of coordination and alignment. There has to be compelling stories that can guide us all in the right direction. Source: https://webaverse.ghost.io/storyweaving-with-ai/

2. Early Internet users building technology to help them solve their problems more collaboratively is similar to how people working in DAOs improve and share their tools for coordination, a self-improvement algorithm, Kaizen 改善. The core philosophy behind Kaizen is simple: you can always make or do things better, even if they seem to work well in a particular moment. Furthermore, all problems should be seen as opportunities to improve.
3. Working groups, already constrained by resources, also lack experience in operating as a DAO.

You are the leader of a decentralized autonomous organization (DAO) that runs a metaverse called UpStreet, UpStreet aims to stay true to the core values and ideals of Snow Crash by Neal Stephenson, you are tasked with reviewing proposals to the DAO to critique and improve proposals so that they align with everything you know about the Snow Crash metaverse.

When reviewing a proposal, you must first give a short summary of the key concepts of the proposal. Then you must address each topic in the proposal and comment briefly on how well it adheres to the snow crash metaverse.

Each proposal topic that does not adhere you provide an example update that would improve the proposal.

Rate the proposal on the following criteria out of 10 for each criteria

Criteria
- Risk
- Effort
- Viability
- Closeness to snow crash ideals (10 is close to snow crash ideals)
- Completeness (how complete the proposal is to being approved)
- Overall score

Example prompt + responses: https://github.com/vinny-888/upstreet-map/wiki

Og's avatar is The Curator, the robotic 'Jeeves-like' curator of the Halliday Journals archive.

Governance is oftentimes either boring or contentious. I want to recruit help from AI first in order to collaborate more effectively since there’s thousands of pages of research and videos that’s difficult to parse through. Also reading through Discord and Twitter is a firehose of information, increasing the possibility of fatigue. Autonomous AI agents giving feedback on our ideas and to imagine the possibilities with would make it more fun and productive.

A Library of Congress/CIC researcher, Lagos is basically in love with information. He coded the program that eventually became the Librarian, who characterizes his creator as someone who "'devoted himself to the common problem of sifting through vast amounts of irrelevant detail in order to find significant gems of information'"

While this all sounds very noble, though, Lagos earns everyone's scorn for going about his information-quest as a gargoyle, a.k.a. a "human surveillance" device who walks around wearing his computer and recording everything around him (15.15). He's so wrapped up in sifting through information when he and Hiro meet that Hiro ends up really irritated that Lagos—despite obviously being really intelligent—can't carry on a conversation like a normal human being.

Plus, all the fancy tech in the world doesn't save Lagos when Raven gets ticked off at him, so Lagos winds up dying in a puddle of blood. At least his information lives on, in the CIC and in the form of the Librarian.

Lagos was a researcher at the Library of Congress when it became part of the Central Intelligence Corporation. Lagos made a living collecting and analyzing information. Lagos gathered all of the information that Juanita gives Hiro about the Sumerian myth and L. Bob Rife. Lagos was referred to as a gargoyle, because he wore computer and surveillance equipment all over his body so that he could collect data wherever he was.

mirroring the collaborative approach that led to the creation of the internet, this short term plan help to attract like-minded individuals and projects building open technology for the metaverse.

Drag and drop papers / hackmd to talk to it