Taking a Leap Forw-AI-rd

The Weekly Round-Up of Everything New in AI

Hey there 👋

Welcome to another exciting edition of Unwind AI! We have a jam-packed edition for you this week with exciting new developments in the world of Artificial Intelligence and we can't wait to share all the juicy details with you. As always, we are dedicated to bringing you the latest and greatest information in AI, and with over 150 new subscribers this week, our community is growing stronger every day. We're thrilled to have you on board!

So buckle up and let's get started!

This issue covers:

1. Latest Developments 🌍

2. News from the Industry 🧑‍🏫

3. Tools of the Trade ⚒️

4. Hot Takes 🔥

5. AI Meme of the Week 🤡

Latest Developments 🌍

No Lights, No Cameras, All Action 🎥

Runway has launched Gen-1, a next-gen generative AI system that can turn existing videos into new, compelling pieces of footage using images or text prompts. It is based on a structure and content-guided video diffusion model trained on images and videos and offers fine-grained control over output characteristics. The platform offers several modes for stylization, storyboarding, masking, and customization, making it a powerful tool for creative professionals looking to experiment with the future of storytelling.

Source: research.runwayml.com/gen1

Dreamy Videos with Dreamix 📹

Dreamix by Google is a video editor that can create new videos based on image and text inputs while maintaining fidelity to color, posture, object size, and camera pose. It is the first diffusion-based method that performs text-based motion and appearance editing for general videos. The approach combines low-resolution information from the original video with synthesized high-resolution information aligned with the text prompt. The method can also be used to animate images and generate subject-driven videos.

Source: dreamix-video-editing.github.io

Google’s Harmonic Helper 🎹

Google has developed SingSong, which generates instrumental music to accompany vocals, making it easier for musicians and non-musicians to create music with their voice. It uses cutting-edge technology to separate vocals and instrumentals from a large corpus of music audio, and trains a model called AudioLM to generate instrumentals based on the vocals. The system has been improved by exploring different ways of representing vocal inputs, leading to a 53% improvement in performance with isolated vocals.

Source: storage.googleapis.com/sing-song

Primer for Biomedical Excellence ⚕️

Microsoft in its latest research paper, introduced BioGPT, a pre-trained Transformer language model focused on the biomedical domain. BioGPT outperforms previous models in several biomedical natural language processing tasks, including relation extraction and QA, and can also generate fluent descriptions for biomedical terms. BioGPT has achieved human parity and has the potential to aid biologists in scientific discovery.

Source: Microsoft Research

Excel-lent Assistance 🔢

Microsoft has developed FLAME, a T5-based model trained on Excel formulas designed to assist with spreadsheet formulas. It is trained using a curated dataset, an Excel-specific formula tokenizer, and domain-specific pretraining objectives. It outperforms larger models like Codex-Davinci, Codex-Cushman, and CodeT5 in 6 out of 10 settings in formula repair, formula auto-completion, and syntax reconstruction tasks.

Source: arxiv.org

ChatGPT but FREE and OpenSource 😲

Open Assistant is an opensource conversational AI developed by LAION, designed to be accessible to everyone. It is a chat-based assistant that can understand tasks, interact with third-party systems, and retrieve information dynamically. The aim is to build the assistant of the future with capabilities beyond writing emails and covering letters, such as meaningful work, API usage, dynamic research, and more. The assistant will be personalizable and extendable by anyone while also being small and efficient enough to run on consumer hardware.

Your Own News Curator 🗞️

Instagram founders have joined hands again to develop Artifact, a personalized news feed app that uses AI to curate articles and stories for its users. The app opens to a feed of popular articles from a list of publishers, ranging from well-known news organizations to small-scale blogs. By tapping on articles of interest, Artifact serves similar posts and stories, much like TikTok's algorithmic predictions, regardless of a user's friends or followers. It is currently in private beta.

(Source: The Verge)

---

News from the Industry 🧑‍🏫

Microsoft Bing + OpenAI = New Search Experience 🌐

Microsoft’s new AI-powered Bing search engine and Edge browser are designed to improve search results and offer complete answers. The new Bing features an improved version of the familiar search experience, a new chat experience for complex searches, the ability to generate content, and a summary of information in the Edge browser. It runs on a next-generation OpenAI’s large language model that is more powerful than ChatGPT called “Prometheus”.

Will it be able to challenge Google’s legacy? Well, we’ll find that soon!

Here’s a sneak peek into the new search engine with Microsoft’s CEO, Satya Nadella.

Google’s Comeback with Bard😎

Google has released Bard, a conversational AI platform powered by LaMDA (Language Model for Dialogue Applications), to provide fresh, high-quality responses by drawing information from the web. Google is also working on integrating AI into its everyday products, starting with Search, to provide deeper understanding and insights into information. It will soon start onboarding individual developers, creators, and enterprises to try its Generative Language API, powered by LaMDA.

Source: Google blogs

Microsoft v/s Google is going to get fierce.🔥

The Competition is not limited to Search 🏇

Google has invested $300 million in Anthropic, a start-up founded by former OpenAI employees, for a 10% stake in the company and will collaborate with Anthropic in developing AI computing systems. Anthropic is also developing its own chatbot, Claude, which is seen as a potential rival to ChatGPT.

Source: golem.de

(Source: The Verge)

Sight, Search, and Google 👁️

Google is making its search process more visually oriented through the use of AI to understand the information in multiple forms including language and images. The latest update includes the ability to use Lens to "search your screen" on Android globally, allowing you to search photos and videos without leaving the app.

Another new feature is Multisearch, which allows you to search using both images and text at the same time, with the option to search locally by adding "near me."

Source: Google Blogs

ChatGPT’s Unprecedented Surge - 100M and counting..🔃

OpenAI’s popular chatbot ChatGPT has reached 100 million users just two months after launching, with an unprecedented growth rate for a consumer app, according to UBS analysts. The app had 590 million visits in January from 100 million unique visitors, compared to TikTok's 9 months and Instagram's 2 years to reach the same milestone.

(Source: The Guardian)

A Battle Over Borrowed Photos 🥷

Stock image site Getty Images has filed a lawsuit against Stability AI, alleging that Stability AI stole 12 million of Getty's copyrighted images, captions and metadata to train Stable Diffusion. Getty asks the court to make Stability AI remove the violating images and pay up to $150,000 for each infringed image. The lawsuit also claims that Stable Diffusion contains distorted Getty watermarks and that Stability AI is aware of this but has not modified its model to prevent it.

(Source: Gizmodo)

Bill Gates' Take on AI Impact 🤔

In an interview, Bill Gates, co-founder of Microsoft, spoke about his work with OpenAI and the potential impact of AI on jobs and medicine. Gates was impressed with the work of OpenAI in scaling up large language models and innovative work with reinforcement learning. He considers AI to be one of the four most important milestones in digital technology, along with the PC, the PC with a GUI, and the internet. He believes that repetitive blue-collar and physical jobs will be the first to be impacted by AI, but tasks involving reading and writing fluency are also being impacted.

---

Tools of the Trade ⚒️

Level Up your Brand Game 🚀

Flair AI is an AI-powered design tool for creating branded content. It has a drag-and-drop interface, allowing users to easily add their product photos into the canvas. The tool also provides smart prompt recommendations with image examples and an autocomplete feature to assist users in their design. It is currently in private beta and is offering free trials.

Source: flair.ai

UI/UX on Steroids ⚡

Galileo AI allows the creation of high-fidelity UI designs from simple text descriptions. Trained on thousands of designs, Galileo AI generates complex UI designs in a flash and can also populate designs with AI-generated illustrations and images. The platform leverages large language models to understand the complex context and accurately fill the end-to-end product copy.

Source: usegalileo.ai

Your Pitch-Perfect AI Co-Founder 🧑‍💼

PowerMode uses GPT-3 technology to help startup entrepreneurs with the ideation and pitch process. It serves as an AI co-founder and helps users turn ideas into slides in seconds by using simple text inputs like industry, competitors, features, etc., making it a time-efficient solution for those looking to bring their startup ideas to life.

Dream Decoded by AI 💭

DreamInterpreter AI leverages the power of AI to provide a deeper understanding of your dream and its underlying themes and emotions. Simply describe your dream in plain language, and AI will analyze the description and provide a personalized interpretation.

AI-Way to Name your Business 📛

Namelix is a business name generator that uses AI to create short and brandable names for businesses. It offers a range of name ideas, with the option to filter results based on specific criteria such as length or keywords. The algorithm learns from saved names, providing better recommendations over time. It stands out from other generators by creating short, branded names relevant to the business concept rather than combining dictionary words to form longer names.

Source: Namelix.com

---

Hot Takes 🔥

Compute is all you need 🧪

Greg Brockman @gdb

Kind of crazy that a computing device designed for desktop gaming has become critical to the emergence of super useful AI systems.

Google 🪢 Bing

Shubham Saboo @Saboo_Shubham_

Me before ChatGPT: Let me Google it Me after ChatGPT: Let me Bing it!

JFK Revived 🪦

Bojan Tunguz @tunguz

“Ask not what your chatbot can do for you – ask what you can do for your chatbot.”

---

AI Meme of the Week 🤡

Shubham Saboo @Saboo_Shubham_

All paths to Data Science leads through Data Engineering. Credits: @christianbdata

That’s all for this week!

Will see you next Saturday with more such content. Don’t forget to subscribe and give your feedback below.

BONUS 🎉

Share this newsletter with three other friends and stand a chance to win a signed copy of my book Neural Search - From Prototype to Production with Jina. Winners will be selected on a monthly basis!

🎁 Every paid subscriber will also receive $39 USD worth of learning resources on trending topics like Python, Data Science, Machine Learning, and NLP!