Autonomous Agents: Friend or Foe?
Steve Jobs envisioned ChatGPT, AI Images in real-time, notes into webpages & more!
Greetings everyone,
👋 Deiniol here - founder of AI Suite.
Some people say we have already reached a level of AGI. Why? Fully autonomous self-prompting agents have arrived. AutoGPT is one of them. It has received 50,000 stars on GitHub in under two weeks. Whilst only the beginning, developers can now collaborate to advance the tech. More below.
🎨 + 💻 = NewArt
Do you want to know how Steve Jobs envisioned ChatGPT? If so, you should check out this publication by NewArt Press. It reveals how the visionary founder of Apple predicted the rise of ChatGPT in a speech he gave in 1983, long before anyone else could imagine it. Don’t miss it, NewArt is a fascinating publication that I love.
“(The) next frontier of prompt engineering imo: AutoGPTs”
- Andrej Kaparthy
📣 Highlights
AutoGPT - Tell your AI what you want and watch it think, plan and execute.
AutoGPT is an experimental open-source application and one of the first examples of GPT-4 running fully autonomously. It truly pushes the boundaries of what is possible with AI. It can create self-prompting AI agents that can achieve various goals given by the user. It can also explain its thoughts, reasoning, and criticism for each action it takes.
Benefits:
🌐 - Internet access for searches and information gathering
💾 - Long-Term and Short-Term memory management
🧠 - GPT-4 instances for text generation
🗃️ - File storage and summarisation with GPT-3.5
Limitations:
⚠️ - Not a polished application or product, just an experiment
👎 - May not perform well in complex, real-world business scenarios
💸 - Consumes a lot of tokens and resources from the GPT-4 and GPT-3.5 APIs, which can be costly and limited
🔄 - May get stuck in loops or dead ends when performing tasks. It may not handle errors or surprises well. It may repeat or generate irrelevant text
AutoGPT is not the only autonomous Agent…
Where to access an Autonomous Agent via a friendly UI:
Twitter buys 10,000 GPUs (Bypass paywall here)
Elon Musk reportedly purchased about 10,000 GPUs for a generative AI project within Twitter.
The project involves a large language model (LLM) that could potentially use Twitter's massive library of user tweets to learn natural language patterns.
The exact purpose and application of the project are still unclear, but some possibilities include improving Twitter search or creating novel advertising methods.
The purchase of the GPUs shows that Musk is committed to AI development within the company, despite signing a cautionary letter that called for a six-month pause in the development of powerful AI models.
The GPUs are estimated to cost tens of millions of dollars and will likely be housed in Twitter's Atlanta data center.
Twitter has also recruited AI talent from DeepMind, a subsidiary of Alphabet, to work on the project.
OpenAI’s Consistency Model (this is big)
OpenAI has released the code for their Consistency Model, a new technique for creating images with AI in one shot. Allowing AI to create images from natural language in real time.
Unlike Diffusion, which needs many steps to remove noise from images, this method can generate realistic images in a single step.
They can also perform zero-shot image editing, which means they can modify existing images without any additional training or data. For example, they can change the colour or the style of an image.
Possible use cases:
Real time editing
NeRF rendering
Real time game rendering
No demo yet although a significant speed increase for image generation is always a winner. We’ve moved from dial up to broadband. How suite 😎
⚙️ Tool 1
Gamma | Beautiful, engaging content with none of the formatting and design work.
Imagine if you could turn your notes into beautiful docs, decks or webpages in a snap. That’s what Gamma does.
It’s an AI-powered tool that simplifies content creation. You just write your thoughts and Gamma creates a gorgeous template for you. No need to worry about design.
Gamma helps you create content that is crisp, visual, and interactive. And it works on any device.
🌟 Other Benefits:
🎉 No more slide masters or template lock-in
🎨 Restyle your entire deck in just one click
⏰ Use a flexible template to work faster
🔗 Embed anything
🎤 Present Live
📊 Measure engagement using built-in analytics
🔎 Where can you try?
Give Gamma a try for free and see the magic. It’s available now: Gamma.
⚙️ Tool 2
Multi·On (beta) | A ChatGPT-powered Web Co-Pilot & AI personal assistant
How to make your browser do anything you want.
Imagine if you could just tell your browser what you want and it would do it for you. That’s what Multi-On does.
It’s like having a personal assistant in your Chrome. You can ask it to book flights, order food, research topics, manage your email, and more. And you can give it multiple instructions/prompts at once. It’s also free.
It’s launching soon on the Chrome Store and there will be a ChatGPT Plugin launching this weekend. Don't miss it.
🧑🏫 Demo/Tutorial
Chaos Prompting: A How-To Guide | Blac.ai
Do you want to collaborate with a powerful and unpredictable partner? If you answered yes, then this process tutorial is for you. Blac.ai shows you how to harness AI to create, shape, and master CHAOS; break free from routines, spark your imagination and work side by side with AI to produce amazing pieces.
🛠 Tools:
🌃 Midjouney /blend
💬 ChatGPT
The below steps are word for word by Blac.ai
1️⃣: Let's create Chaos. Go to Chat-GPT, and ask for something like the following. Here's how I got a solid prompt, starting with: "give me a completely random string of words that are descriptors for a scene" which leads to a full scale prompt.
2️⃣: Let's guide Chaos. I put this directly into Midjourney, V5, only adding --s 1000 + 2:3 Full prompt: "Electric clouds levitating over twinkling neon lights, while bustling pedestrian traffic and distant sirens provide a pulsating shadow to suspended time..."
3️⃣: Now... W E B L E N D . I want something more 'street' centric, so I ran a new set of images from another prompt + blend mash (1) and blended it with one of the results from our Chaos prompt (2)
✅ annnnnnnd here's the result I'm the happiest with! 👇
This allowed me to create in a way I've never done before and produce a finished product that I would not have discovered otherwise. More importantly, it still has a piece of me in it, my expression and my direction/guidance.
If you found value in this week’s newsletter, please support us by liking our publication at the bottom. It helps a lot 🤗
Until next week, earthlings, bots and cyborgs, code out!
Deiniol from Sydney (not Bing) 🇦🇺
Reply back to this email with your Twitter handle (or just say 👋 and I will provide an exclusive 50% discount to WriteSonic :)
🏆 Sponsor Us
Get your business, podcast or AI Art in front of thousands of email subscribers, Twitter followers and AI enthusiasts, startup founders, aspiring entrepreneurs and artificial intelligence students who visit our website frequently.