Gemini Ultra is coming!

PLUS: The open-source (and free) GPT Store competitor

Welcome, AI enthusiasts.

A new leak just revealed a potential makeover for Google Bard – AND the launch of Google’s top-tier Gemini Ultra model.

If the rumors are true, we could be in for a wild week! Let’s investigate…

In today’s AI rundown:

  • Leaked doc reveals Bard rebrand and Gemini Ultra launch

  • Hugging Face launches open source assistant creator

  • How to generate 3D assets from text with Meshy

  • AI model learns human language from baby

  • 6 new AI tools & 4 new AI jobs

  • More AI & tech news

Read time: 4 minutes

LATEST DEVELOPMENTS

GOOGLE

Image source: Dylan Roussel (@evowizz on X)

The Rundown: A leaked changelog circulating on social media suggests Google is preparing a major revamp of its Bard chatbot under the new name Gemini, alongside the launch of its highly anticipated Gemini Ultra model.

The details:

  • The leaked changelog shows Bard being rebranded as Gemini on Feb. 7th in an effort to better align the brand with its underlying generative models.

  • A paid ‘Gemini Advanced’ tier, built on the more powerful Ultra model, is also listed for a Feb. 7 launch.

  • Gemini will introduce its first Android app on select Pixel and Galaxy devices, expanding globally over time.

  • Canada also (finally) gets Gemini access on the web, with App access coming soon.

Why it matters: Given previous rumblings of phasing Bard out for Gemini, the rebranding news isn’t shocking to hear. More exciting is the launch of Ultra — and if its capabilities live up to the hype, the model could re-ignite the LLM acceleration race and push OpenAI for a faster GPT-5 rollout.

TOGETHER WITH ARTISAN

The Rundown: Meet Ava – your dedicated AI sales assistant. Onboard Ava with a simple 10-minute chat, after which she will start finding leads & writing emails 24/7.

Ava is the first Artisan, able to:

  • Find targeted sales prospects’ information from a database of 270M+

  • Write & send hyper-personalized emails

  • Autonomously schedule meetings in your calendar

Hire Ava today to get more sales meetings, without hiring a human BDR.

HUGGING FACE

Image source: Hugging Face

The Rundown: Hugging Face just launched a free tool for easily building customizable AI assistants powered by open source models, positioning it as an alternative to OpenAI’s GPT Store offerings.

The details:

  • Users can create assistants in ‘2 clicks’, with the ability to choose between various open source models like Claude, LLaMA 2, and Mixtral.

  • Assistants are free to create, shareable and browsable via a central repository, similar to OpenAI's GPT Store.

  • Unlike GPTs, assistants do not support RAG, web search, actions, and image generation, though the features are on the roadmap for future integration.

Why it matters: While still missing some key features, HF’s assistants are a huge step for open source AI — and the ability to utilize different LLMs will enable even more creative applications. The tool is also yet another example of open source tech quickly catching up to industry leaders.

AI TRAINING

The Rundown: Meshy has recently unveiled its new Text to 3D tool, Meshy-2. It enables users to create high-quality 3D models quickly.

Here is a quick step-by-step:

  1. Sign up: Head to Meshy to create an account for free.

  2. Create: Opt for Text to 3D BETA, describe your model, and pick an art style.

  3. Refine: Meshy generates 4 preview models; choose one or more to refine for more detail.

  4. Download: Click the right-side download button to save your 3D assets in various file formats.

AI RESEARCH

Image source: Wai Keen Vong

The Rundown: An AI model learned to recognize words like "crib" and "ball" by training on first-person video and audio captures from a baby's life, offering insights into human language development in a new study from researchers at New York University.

The details:

  • Researchers gathered 61 hours of helmet cam footage from an Australian baby to show the infant's perspective.

  • The AI neural network was exposed to 250,000 word-image pairs from the baby's activities to learn associations.

  • In evaluations, the AI successfully matched words to images 62% of the time, comparable to other models trained on vastly more data.

  • The approach challenges theories that innate knowledge is needed to learn language, showing general learning skills may be just as valuable.

Why it matters: This fascinating research could challenge traditional theories on how we acquire and learn language — as well as help to adapt AI models for understanding the physical world more intuitively rather than relying on huge datasets.

NEW TOOLS & JOBS

  • 📸 Kondor for iOS- ChatGPT for your phone camera (link)*

  • 🎥 Pictory GPT- Effortlessly create captivating videos (link)

  • 📊 Matrices- Simplify data analysis and automate tasks for better insights (link)

  • 📚 Books GPT- Your personal AI guide for discovering and recommending books (link)

  • 🛠️ Promptly- Build generative AI applications without coding experience (link)

  • 🌐 TereraAI Learn Language GPT- Learn a new language interactively with adaptive scenarios (link)

  • 🔧 Open AI- Reliability Engineer, Research Platform (link)

  • 🤗 Hugging Face- Machine Learning Engineer Internship (link)

  • 👨‍💼 Anthropic- Payroll Manager (link)

  • 🔍 Cohere- Engineering Program Manager, Search & Embedding ML (link)

*Sponsored listing

QUICK HITS

A new study revealed that AI agents, including advanced LLMs like GPT-4, can potentially escalate military and diplomatic tensions in simulated war games.

A mysterious ‘DeluxeChat’ model is making waves on Imsys Arena, with impressive performance expected to be in line with top available models.

Kleiner Perkins VCs Mamoon Hamid and Ilya Fushman estimated that ‘more than 80 percent’ of pitches they hear now involve AI in an interview with TechCrunch.

Apple CEO Tim Cook hinted at a significant AI announcement coming from the company later this year during an analyst call following earnings reports.

Google released MobileDiffusion, a new approach that allows for near-instantaneous text-to-image generation on mobile devices.

Alibaba’s Quen-VL-Plus is quietly gaining ground on the top multimodal LLMs, with its vision capabilities reportedly matching or exceeding GPT-4.

AI lobbying has reportedly surged over the last year, skyrocketing by 185% in 2023 with over 450 organizations engaging in response to increasing calls for AI regulation.

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.

Login or Subscribe to participate in polls.

If you have specific feedback or anything interesting you’d like to share, please let us know by replying to this email.

Reply

or to participate.