- The Rundown AI
- Posts
- Google goes open-source!
Welcome, AI enthusiasts.
Google is keeping its foot on the AI gas pedal.
The tech giant just dropped a small (but powerful) open-source model called Gemma — ironically leaving OpenAI as the least ‘open’ major player left. Let’s dig in…
In today’s AI rundown:
Google releases new open-source models
Magic claims major AI reasoning breakthrough
Summarize PDFs with Adobe Acrobat AI
5 new AI tools & 4 new AI jobs
More AI & tech news
Read time: 3 minutes
LATEST DEVELOPMENTS
Image source: Google
The Rundown: Google just unveiled Gemma, a new open-source family of AI models, allowing developers to build freely on language capabilities from its proprietary Gemini without restrictions.
The details:
Gemma comes in 2B and 7B model sizes, outperforming models like Mistral and LLaMa 2 on key benchmarks.
While not as powerful as Gemini, the Gemma models offer speed and cost efficiencies — with the ability to run directly on a laptop.
Gemma is commercially available on HuggingFace, Kaggle, and Google’s Vertex AI.
Why it matters: While not as flashy as Gemini Ultra, Gemma’s small but mighty capabilities lend well to on-device use cases, allowing builders to get creative. Google’s open-source turn may now put pressure on OpenAI, who ironically stands as one of the only major AI players with fully ‘closed’ models.
TOGETHER WITH IMAGINE AI LIVE
The Rundown: Join us at the Imagine AI Live conference, March 27-28, at the new Fontainebleau Las Vegas, and discover how AI is transforming the business landscape.
Meet leaders of the generative AI revolution like:
Mark Heaps, Chief Technology Evangelist at Groq Inc.
Bindu Reddy, CEO of Abacus AI
Div Garg, Founder of MultiOn
Spots are filling up fast! Don't miss out on this chance to future-proof your business. Sign up before February 22nd, and save 10%!
MAGIC
Image source: The Information
The Rundown: A new report just revealed that Magic, a startup building AI coding “coworkers”, claims to have made a massive AI breakthrough in context length and reasoning ability.
The details:
Magic claims its model now can ingest up to 3.5M words (nearly 5x more than Google's latest Gemini 1.5 update).
The startup also reportedly made an ‘active reasoning’ breakthrough reminiscent of OpenAI's rumored Q* discovery.
Last week, the company announced a $117M raise led by former GitHub CEO Nat Friedman.
Magic has already lured top GitHub talent like its VP of design, with speculation it will ship a huge AI product in the next few months.
Why it matters: These claims shed more light on last week’s large funding round — and with recent major jumps in context lengths, we’re getting to near-infinite levels of processing ability. OpenAI’s mysterious Q* discovery could also return to the spotlight if Magic made a similar breakthrough.
AI TRAINING
The Rundown: Adobe Acrobat just released an AI assistant that will analyze, summarize, and allow you to ask questions about PDFs. This tool will save you a ton of time and help you learn any subject.
Here’s a quick step-by-step:
Head to Adobe Assistant, login, and sign up for a free trial or subscription.
Upload your PDF, and the AI Assistant will provide a brief overview of the document.
You can now pose questions through the chatbot or use the prompts suggested by the assistant to comprehend the document's content quickly.
Try chatting with lengthy papers, business outlines, books, and more!
NEW TOOLS & JOBS
🧠 GigaBrain- Discover real insights from Reddit discussions with curated content (link)
🌐 Rio- Discover overlooked global news, always stay informed (link)
🎉 Linkedin Post Formatter GPT- Format LinkedIn posts for maximum engagement (link)
🤖 GovernGPT- Automate investor updates and streamline asset management workflows (link)
🍈 Melon- Your AI thought partner and second brain (link)
QUICK HITS
Section is holding a free event, Where's the Money in AI?, today (Thursday) at 1 p.m. ET. Learn from Exponential View's Azeem Azhar, where the ROI is in AI projects and business models. RSVP here.*
Nvidia blew past expectations with its Q4 earnings report, with a 265% year-over-year increase and massive data center revenue pushing the company’s valuation above $1.5T.
Google announced that Gemini capabilities are now available across Workspace apps like Gmail, Docs, and Slides for Premium users.
Chinese startup Moonshot AI raised over $1B from backers, including Alibaba, for its development of LLMs handling long-form text.
Scale AI has been tapped by the Pentagon to develop a framework for testing and evaluating LLMs, aiming to safely deploy AI in military applications.
Google’s Gemini is drawing viral criticism after users have struggled to generate images of Caucasians despite a variety of prompts.
ChatGPT briefly went off the rails, hallucinating incoherent (and often hilarious) responses before OpenAI announced the issue had been resolved.
*Sponsored listing
THAT’S A WRAP
SPONSOR US
Get your product in front of over 450k+ AI enthusiasts
Our newsletter is read by thousands of tech professionals, investors, engineers, managers, and business owners around the world. Get in touch today.
FEEDBACK
How would you rate today's newsletter?Vote below to help us improve the newsletter for you. |
If you have specific feedback or anything interesting you’d like to share, please let us know by replying to this email.
Join the conversation