Ai ✧ flavour of qual ✧ iA

Making AI Talk: From Computer Language to Human Though Prompting AI can be tricky. There are many cool approaches to it, but not all of them work for everything, and they can be hard to understand. But what if we could make it easier and more useful? This is about changing how we use AI. Right now, AI uses small units of information called 'tokens' to think and respond. What if we could teach these tokens to think like humans? Role-playing is a good way to do this. When we ask AI to act like a real person, it learns to think and respond like that person would. This makes AI more accurate and useful. However, there's a catch. AI often follows the path it starts on. If we start with wrong or misleading information, AI might keep going in that direction. Using mental models To use role-playing well, we can use 'mental models' or ways of thinking. For instance, we could ask AI to "Think like Elon Musk using the 'First Principles' model". This tells AI to break down a problem to its simplest parts, just like Elon Musk would. Or if we need to make a quick decision, we could ask AI to "Use the 'OODA Loop' model like a military strategist". This tells AI to watch the situation, think about the new information, decide what to do next, and then do it. Modeling thinking This way of using AI is new and different. We're not just asking AI to give answers. We're teaching it to think and help solve problems. This could make AI even more useful and bring us closer to understanding how it works. Prompts used (you need to paste both) 1. Generating experts (replace area and problem with yours) 2. Providing instructions on how to think Other ideas: you can tweak the prompt the way you want or introduce several experts at a time to get more ideas on your problem. You can ask experts to ask you more questions about your goals, on why do you want to solve this task, etc. Prompting LLM isn't about discarding thinking. It's about making thinking more effective and providing you with more feedback. Example in ChatGPT

🚀 You won't believe this! Researchers from NVIDIA and Stanford have created a crazy smart Minecraft agent called Voyager, and it's learning all on its own, powered by none other than GPT-4! Here's the rundown: 1️⃣ Voyager's got this cool thing called an "automatic curriculum" that keeps it exploring non-stop. 2️⃣ It's building a massive library of skills, teaching itself how to do all sorts of complex tasks in the game. 3️⃣ And it uses this nifty "iterative prompting mechanism" to learn from its mistakes and get better over time. And the best part? It doesn't need any fine-tuning! Voyager just asks GPT-4 questions, and off it goes. Now, here's where it gets wild. In tests, Voyager was an absolute beast! It found 3.3x more unique items, traveled 2.3x longer distances, and got through key tech tree milestones up to 15.3x faster than anything we've seen before. But wait, there's more! When dropped into a brand-new Minecraft world, Voyager took everything it learned and solved new tasks from scratch. It's like the ultimate Minecraft survivalist! So, there you have it. We're living in a world where AI can learn to master Minecraft all by itself. What's next, right? 🤖👾🎮

Little comparison of version 3 and version 5, more examples in the comments

MidJourney v5 blend ability is incredible All what you need to know

GPT-4 https://openai.com/research/gpt-4 -> додаткові посилання буду додавати в коментарі

Boom! ChatGPT API + Whisper API The new Chat API calls gpt-3.5-turbo, the same model used in the ChatGPT product. It’s also our best model for many non-chat use cases; we’ve seen early testers migrate from text-davinci-003 to gpt-3.5-turbo with only a small amount of adjustment needed to their prompts. Learn more about the Chat API in our documentation. Pricing It’s priced at $0.002 per 1K tokens, which is 10x cheaper than the existing GPT-3.5 models. Whisper API The API is priced at $0.006 / minute, rounded up to the nearest second. https://openai.com/blog/introducing-chatgpt-and-whisper-apis

[✧] GPT: Here we go again ▶️ soundtrack for the post ——— [✧✧] Integrations - Raycast (Link) [Integration of OpenAI into Raycast app] - Notion (Link) [Notion has made the GPT service available to their service] ——— [✧✧] Semantic Search - Metaphor (Link) [Semantic searching with unusual results] - Perplexity (Link) (Chrome ext.) [Search engine while browsing the web] [Search and query on the content of the page/document] - Phind (Link) [The AI search engine for developers] - arXiv Explorer (Link) [Semantic search engine for ArXiv] - Andi Search (Link) [Chat for search] - Kagi Search Engine (Link) [Beta tested contextual search with GPT] [In February 2023 it will be implemented in the Kagi search engine] - CrowdView (Link) [Search content from Forums] - Kaito (Link) [Search in the area of Web3 - crypto and blockchains] [Waitlist] - You Search Engine (Link) [The AI Search Engine] ——— [✧✧] Semantic Research - Elicit (Link) [Ask a research question and get insights from a research papers] - Researchrabbit (Link) [Surf through the research papers] [not a GPT] (Link) - Consensus (Link) [Ask a question, get conclusions from research papers] - Omniscience (Link) [All-in-one AI platform for research] - Explainpaper (Link) [Get help from the GPT while you are reading a research paper] - Humata (Link) [Ask GPT based on your uploaded files] ——— [✧✧] Chrome plugins with GPT - writeGPT (Link) - CasperAI (Link) - Cmd+J (Link) ——— [✧✧] Newsletters - ⭐️ Ben's Bites (Link) (Bot) — @Ethea

[Reddit] • Prompt itself. #ChatGPT