Anthropic's Context Protocol, 1000 Agents in Minecraft, and Amazon’s AI

PLUS HOT AI Tools & Tutorials

Nov 29, 2024

Black Friday Sale 🔥

Grab your 20% Discount on Creators AI Subscription. Offer Expires on Sunday!

Hello and welcome to our weekly roundup!

Today, we discuss a new standard to strengthen AI integration among enterprises, a new rival for OpenAI's best model, how 1000 LLM agents are doing in Minecraft, and more!

We've got a ton of news, tools, and guides piled up, so let's get to it.

This Creators’ AI Edition:

Featured Materials 🎟️
News of the week 🌍
Useful tools ⚒️
Weekly Guides 📕
AI Meme of the Week 🤡
AI Tweet of the Week 🐦
(Bonus) Materials 🎁

Your advertisement could be featured here!

Sponsor a spot in our newsletter to connect with subscribers interested in AI, technology, and startups. For inquiries, click the button below.

Featured Material 🎟️

Anthropic’s Model Context Protocol

An abstract illustration of critical context connecting to a central hub

An important release from Anthropic that hasn't gotten enough attention. The company has proposed a new standard for connecting AI-based assistants to systems where data is stored. Model Context Protocol (MCP) platform will help AI models produce better and more relevant responses to queries. This should be especially useful for enterprises that are actively integrating algorithms into their daily tasks.

Using MCPs, AI models can extract data from sources such as business tools and task software, content repositories, and application development environments. This allows developers to create two-way connections between data sources and AI-based applications (e.g., chatbots). The best part is that MCP is an open standard.

This means that the system works not only with Anthropic's Claude family but also with other models like GPT and Gemini.

The three main components of the Model Context Protocol are already available:

The Model Context Protocol specification and SDKs
Local MCP server support in the Claude Desktop apps
An open-source repository of MCP servers

Current Claude for Work customers can begin testing MCP servers locally by connecting Claude to internal systems and datasets. Anthropic will also soon provide developer toolkits for deploying remote production MCP servers.

Keep your mailbox updated with key knowledge & news from the AI industry

Anthropic noted that companies such as Block and Apollo have already integrated MCP into their systems, while tooling companies, including Replit, Codeium, and Sourcegraph, are gradually adding MCP support to their platforms. The developer proposes turning MCP into a collaborative project and open-source ecosystem, so he invites everyone to try out the new solution and share feedback.

On paper, Model Context Protocol looks great. However, I see a few points worth voicing here. First, Anthropic hasn't provided any real tests demonstrating a significant benefit to organizations using MCP. And second, it's pretty hard to say whether OpenAI, Google and others would agree to promote a direct competitor's system. Personally, I highly doubt it.

Grab your 20% Discount on Creators AI Subscription. Expires on Sunday!
Use this link https://thecreatorsai.com/bfcm24

News Of The Week 🌍

Alibaba Releases Its Reasoning Model

Alibaba has released QwQ-32B-Preview, a new reasoning AI model that challenges OpenAI's o1 series. Here are its key features:

32.5B parameters, allowing it to process prompts up to 32,000 words in length.
It outperforms OpenAI's o1-preview and o1-mini models on specific benchmarks, such as the AIME and MATH tests, which evaluate logic puzzles and math problem-solving abilities.
The new model incorporates self-verification capabilities, enabling it to fact-check itself and reduce errors, albeit at the cost of increased solution time.

QwQ-32B-Preview is already available for download under the Apache 2.0 license, allowing commercial use, though only certain components are released. This model looks very impressive, although it does have some limitations. In particular, it showed unexpected language switches, looping, and underperformance in common-sense reasoning tasks. The model's responses also align with Chinese regulatory standards, avoiding politically sensitive topics.

Brave Search Gets AI Chat Support

Brave Search has introduced a new chat mode feature, allowing users to ask follow-up questions regarding their initial queries. The company says the feature builds on its existing Answer with AI tool, which generates more than 11 million daily answers. The idea is to allow users to explore any topic in detail by talking to the AI rather than stopping at a single query. The feature is free and can be accessed via the address bar of the Brave browser or any web browser via the Brave Search website.

By the way, the other day we were just discussing AI Search and how you can use it to promote your projects:
AI Search Optimization: How to Get Ranked by AI
Stepan Ikaev
·
Nov 28
Read full story

The Information: Amazon Develops Video AI Model

According to The Information, Amazon is developing a generative AI to process text, images, and video. This platform is built on a new big language model codenamed Olympus. Insiders speculate that with this AI, Amazon wants to reduce its own dependence on its main partner in this field, Anthropic. Olympus is expected to be released as early as next week at the annual AWS re:Invent conference.

If The Information's insider is confirmed, Amazon will finally close out the big-list of companies that have entered the AI market with their own platform.

Sharing is caring! Refer someone who recently started a learning journey in AI. Make them more productive and earn rewards!

Refer a friend

1000 LLM Agents Built a Community in Minecraft

Finally, important experiments using AI. Startup Altera sent 1,000 LLM agents with a master's degree in law to a Minecraft server and showed what came out of it. As MIT Technology Review reports, given just a nudge via text prompts, they developed a remarkable range of personality traits, preferences, and specialized roles without further input from their human creators.

In fact, the agents could construct several variations of their society with hierarchies, evaluate each other, and take different roles.

Beyond jokes, Altera's experiment is a really important initiative. This is how the startup decided to evaluate how effectively AI agents will work with each other in the real world. As progress is made, automation will spread more and more and it is better to prepare for the moment when different models collide in advance. I suggest reading the full story.

ElevenLabs’ New Feature Lets You Create Podcasts

ElevenLabs has introduced a feature to create podcasts from text with two AI hosts. GenFM allows you to voice texts from PDFs, e-books, and documents with AI voices in 32 languages. The user can also choose different intonations and accents. Theoretically, this will be particularly useful as an educational tool: any boring report or lecture can now be turned into a podcast for background listening. The feature is free in the startup's ElevenReader app on iOS and Android.

Creating AI podcasts is becoming more accessible. In September, Google revealed NotebookLM, a similar feature for generating podcasts based on text notes. However, unlike ElevenLabs, it only supports English.

OpenAI Sora Video Generator Leaked Online

OpenAI's Sora video text-to-speech generator was briefly leaked online by a group of early testers protesting what they call “art stimulation” and the exploitation of creative professionals. The leak occurred on Nov. 26, when the group posted the project on Hugging Face, giving public access to Sora for about three hours before OpenAI shut it down.

During the short access period, users could create 10-second videos at resolutions up to 1080p. After Sora was removed from Hugging Face, OpenAI temporarily suspended all users' access to the model while it investigated.

Artists also claimed that they were pressured to promote Sora as a useful tool for artists without proper compensation. However, this story doesn't have any evidence or corroboration right now, so I remind you to be skeptical.

Useful Tools ⚒️

Expression Editor - Edit Facial Expression with Confidence using AI

Perplexity Shopping – The fastest way from search to purchase

Canvas by MindPal – Canvas to run AI agents & multi-agent systems

Cogent – If Google Drive met Quizlet and ChatGPT

Agentplace – The platform for creating interactive AI websites and apps

Agentplace is a recently launched tool designed to build AI-based interactive platforms. With its help, you can build a website or app without coding skills. The entire process requires only prompts, which are processed through ChatGPT and turned into a ready-to-deploy product. The finished site or app will also support voice mode, dynamic UI, and common sense (AI understands context and handles unexpected situations intelligently.)