Google goes all-in on AI
Google unveiled a bunch of new things in AI on the last Google's mind-blowing I/O event.
Hey there, tech-savvy chefs! Welcome to The AI Kitchen, your trusty partner in crime when it comes to all things AI.
We're like the secret sauce to your burger, the marshmallows to your hot cocoa, and the AI assistant to your burning curiosities. Just picture us as your own personal Jarvis, ready to serve you the latest and tastiest trends in the world of artificial intelligence.
So grab your spatula (or maybe just your coding keyboard), put on your AI apron, and let's whip up some serious AI magic together!
"Hey Alexa, what's cookin' in the world of AI today?"
Small Claude AI revolution: 100k tokens in one single prompt
New OpenAI's model: Text-to-3D
ImageBind: new multi-modal embedding by MetaAI
Google goes all-in on AI
Let's dive in.
Small Claude's revolution: 100k tokens in one single prompt
Anthropic, the AI research startup founded by former OpenAI members, brings us some AI wizardry!
Anthropic has upgraded Claude, their text-generating AI model, by expanding its context window from a measly 9,000 tokens to a mind-boggling 100,000 tokens. That's right, they've leveled up!
Say goodbye to those pesky "The message you submitted was too long" errors. Claude's enhanced context window allows it to analyze not just a few pages but hundreds of them. It's like giving your AI a turbo boost!
Need some perspective? Claude can digest and analyze the entire text of "The Great Gatsby" in less than a minute. Talk about speed-reading on steroids!
This leap forward leaves OpenAI's GPT-4 in the dust with its puny 32k context window (currently in beta). Claude is the new champ, my friends.
So, how can you get your hands on Claude's upgraded powers? Well, for now, it's only available to Anthropic's business partners. But hey, you can always try your luck and request API access on their website. Give it a shot, who knows?
New OpenAI's model: Text-to-3D
OpenAI is back in action with another mind-blowing project. Brace yourselves for some text-to-3D magic with their latest creation: Shap-E!
OpenAI has just unleashed Shap-E, their text-to-3D model, along with the research and code for all you tech enthusiasts out there. Get ready to dive into the world of three-dimensional awesomeness!
Shap-E takes things to a whole new level by producing complex and diverse 3D assets in record time. It's faster, more efficient, and offers better sample quality compared to its predecessor, Point-E. OpenAI is stepping up their game!
Currently, Shap-E has some limitations. It's limited to single object prompts with simple attributes, resulting in pixelated and rough outputs. But hey, don't lose hope! With the lightning-fast pace of AI development, we might just see text-to-3D printers become a reality sooner than you think. Imagine printing your own dreams!
OpenAI is constantly pushing boundaries and raising the bar for what's possible. Keep an eye out for more mind-bending projects coming your way.
ImageBind: new multi-modal embedding by MetaAI
Meta AI has unveiled their groundbreaking multi-modal embedding that takes things to a whole new level. It doesn't just work with text, images, and videos, but also depth, infrared, and even the object's movement and position. Talk about a full-spectrum experience!
What are embeddings, you ask? They're like superpowers for machines, helping them make sense of complex data. For instance, OpenAI's text embeddings measure similarity between text strings, enabling AI-powered search, grouping, recommendations, and long-term memory. It's all about understanding and connecting the dots.
Say hello to ImageBind, the multi-modal embedding that can measure the similarity of all different types of data. We're talking about the ultimate tool for rich multimedia search. Think of it as the Google AI search on steroids, capable of handling all kinds of media.
The use cases for multi-modal embedding are endless. Brace yourselves for some mind-bending possibilities:
Rich multimedia search: Imagine searching for information across various types of media, from images to videos and more. Meta AI is taking search to a whole new dimension!
Sentiment analysis: By analyzing facial expressions and combining them with spoken words, you can unlock a deeper understanding of sentiment. It's like reading emotions with AI!
E-commerce analysis on steroids: Dive into the world of buyer behavior by analyzing product images along with other data. Get ready to supercharge your e-commerce game!
Meta AI is breaking barriers and opening up a realm of possibilities with their multi-modal embedding.
Google goes all-in on AI
Google unveiled a bunch of new things in AI on the last Google's mind-blowing I/O event.
Here are takeaways from the event:
Duet AI for Google Workspace: Prepare for some serious enhancements! AI is coming to Gmail, Docs, Sheets, and Slides, taking productivity to the next level.
Google Maps goes immersive: Get ready for an 'Immersive View for Routes' feature. Strap on your virtual seatbelt and embark on a visually stunning journey.
Magic Editor arrives: Google Photos is getting a magical touch with the 'Magic Editor' feature. It's like having a photo editing wizard in your pocket.
Google Search gets chatty: Say goodbye to dull search queries and hello to interactive conversations! Google Search is leveling up by responding to questions in a conversational manner, just like ChatGPT.
Power of PaLM 2: Google's powerful new Language Model (LLM) is set to unleash its magic across more than 25 new Google products.
Bard goes global: Bard is spreading its wings and becoming available in 180 countries.
Adobe Firefly partnership: Get ready for image-to-text magic! Google has partnered with Adobe Firefly to bring image-to-text functionality directly into Bard.
Med-PaLM: Say hello to Med-PaLM, Google Research's Language Model designed specifically for the medical domain. It's like having a brilliant medical AI assistant by your side, ready to help healthcare professionals provide top-notch care.
MusicLM strikes a chord: Ever had a musical idea and wished it could come to life? Well, MusicLM is here to grant your wish! Describe your musical vision, and let the AI bring it to life. Get ready to compose musical masterpieces with the help of AI.
StudioBot, coding buddy extraordinaire: Meet StudioBot, your trusty AI coding buddy. Get ready for some serious coding adventures as StudioBot assists you in unleashing your developer superpowers.
Now, here's a fun fact: During the event, Google dropped the word "AI" a staggering 143 times. They're definitely AI enthusiasts! And guess what? Google's stock increased by a jaw-dropping $56 billion. Talk about making waves!
That's a wrap, folks! Relive the magic by checking out the epic recap videos.
Top-5 AI tools this week
Superdash HQ - Custom chatbot builder on top of chatGPT API (link)
Instascribe - Supercharge your Instagram copy with AI (link)
Transformer Agents from HuggingFace - A natural language API for talking to Transformers and Diffusers (link)
Chatcraft - Developer-focused open source ChatGPT (link)
AI Background Changer - Realistic AI backgrounds generated for your product photos (link)
Learn
Teach-O-Matic - Making instructional "how to videos" on any topic (link)
Fixing LLM hallucinations with retrieval augmentation in LangChain (link)
How to make custom AI chatbots for your law practice trained on your own documents (link)
Weekly meme
Weekly AI image
Weekly Thread
Weekly Midjourney Prompt
cinematic fashion portrait, photo by Peter Lindbergh, young woman model posing in vogue pose 1994 Pirelli Calendar wearing wardrobe by Alexander McQueen surrounded by smoke and dust, exterior with car wrack, wide angle
That's a wrap for today.
See you next week! If you want more tasty AI treats, be sure to follow our AI Chef on Twitter (link).
If you love this episode and want to support us, spread the word about us by sharing The AI Kitchen with friends. We really appreciate it!
Thank you for reading!
What'd you think of today's edition?
Help me to understand what you think about this episode. Just reply with a number (1, 2, or 3) to this email.
1 - Damn good
2 - Meh, do better
3 - You didn't cook it
Your compadre,
Anton "AI Chef" Cherkasov