Navigating through the Innov-AI-tion

Hey there 👋

Welcome to the latest edition of Unwind AI! We are beyond delighted to have over 150 new subscribers join us this week, bringing us closer to our goal every day of keeping everyone informed and up-to-date on the latest advancements in AI.

The past week has been a whirlwind of new developments in the field, and we are eager to share all the exciting news with you. From breakthroughs in natural language processing to new applications of AI, there is a lot to cover in this edition.

So sit back, grab a cup of coffee, and keep reading!

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 🌍

Who wrote it? I or AI? 🤔

OpenAI has created an AI classifier that can distinguish between text written by humans and AI-generated text. The classifier is not 100% accurate, but its reliability improves with longer text and has improved significantly from previous classifiers on recent AI systems. The classifier is being made publicly available for feedback.

OpenAI has truly stolen the thunder this week! First, it lights the fire and is now donating extinguishers!! 🧯

From Text to Tunes 🎼

Google has introduced MusicLM, a model that generates high-quality music based on text descriptions. The model treats the process of music generation as a sequence-to-sequence task. It generates high-quality music that lasts for several minutes and outperforms previous systems in audio quality while adhering to the text description. To support further research, Google has released MusicCaps, a dataset composed of 5.5k music-text pairs with rich text descriptions provided by human experts.

Check out the sample given below. Although cherry-picked, the output is really good!

Text Prompt: “A rising synth is playing an arpeggio with a lot of reverb. It is backed by pads, sub bass line and soft drums. This song is full of synth sounds creating a soothing and adventurous atmosphere. It may be playing at a festival during two songs for a buildup.”

Generated Audio 👇

The Fourth Dimension of Storytelling🎥

Meta has introduced MAV3D, a method for generating 3D dynamic scenes from text descriptions. MAV3D uses a 4D Neural Radiance Field and is optimized for appearance, density, and motion. The approach does not require any 3D or 4D data, and the resulting video can be viewed from any angle and blended into any 3D environment.

Motion to the Beat of Text 🏃‍♂️

T2M-GPT is a method of generating human motion from text descriptions. It uses a Vector Quantised-Variational AutoEncoder (VQ-VAE) combined with GPT to turn the text into motion. It has shown to outperform other approaches and is able to generate motions quickly and accurately. To use T2M-GPT:

1. provide a text description of the motion you want,

2. choose the desired output method (sketch skeleton or SMPL mesh), and

3. Voila! the motion will be generated.

The Next Generation of Conversational Search 🌐

Perplexity AI has announced a major update to Perplexity Ask, a conversational search engine, combining the capabilities of ChatGPT with the latest references pulled from a search engine. The update allows you to read answers with up-to-date sources and ask follow-up questions, seamlessly converting the question into a new search query, eliminating the need to rephrase the question or think of keywords.

News from the Industry 🧑‍🏫

Become a Prompt Engineer 🖌️

If you are interested in learning the next big thing in AI, prompt engineering is the way to go. Enhance your skills and take control of AI's power with Learn Prompting’s beginner-friendly and dynamically updated course on prompt engineering. It is perfect for anyone irrespective of the background, be it tech or non-tech. Enhance your ability to communicate with AI and automate your tasks by learning techniques and instantly applying them. And the best part, it is completely FREE!

OpenAI launches ChatGPT Plus 🍫

OpenAI has released ChatGPT Plus, the new subscription plan for ChatGPT, which is available for $20/month. The plan offers:

• general access to ChatGPT even during peak times,

• faster response times, and

• priority access to new features and improvements.

The plan is currently available in the US but will expand to other countries soon. Free access to ChatGPT will still be offered. Sign up for the waitlist here!

AI-powered Smart Meetings 🧑‍🤝‍🧑

Microsoft Teams Premium is bringing in the latest AI advancements to Team’s offerings with OpenAI’s GPT-3.5 to make meetings more intelligent, personalized, and protected. It offers over 400 new AI-powered features including meeting notes, recommended tasks, personalized highlights, live captions, and translations in 40 spoken languages. Teams Premium is available for $10 per user per month, with a limited-time offer of $7 per user per month.

Zoom-ing into Seamless Customer Support 🕵️

Zoom has launched Zoom Virtual Agent, an AI and chatbot solution that enables businesses to deliver prompt and personalized customer experiences through NLP and machine learning. It interacts with customers 24/7 on multiple support channels, integrates with CRM, chat, and contact center platforms, and requires minimal maintenance.

Google’s version of ChatGPT 📲

Google had introduced LaMDA (Language Model for Dialog Applications), an open-domain large language model, that can converse on any topic and respond to anything you ask it. Google is now providing access to this model via AI test kitchen to early testers on a rolling basis. Register your interest here.

Unveiling the Privacy Dilemma 🔏

A research demonstrates that diffusion models, such as DALL-E 2, Imagen, and Stable Diffusion, are vulnerable to privacy issues due to the fact that they memorize individual images from their training data and emit them at generation time. As a result, the authors developed a generate-and-filter pipeline to extract training examples from state-of-the-art models. They trained hundreds of diffusion models in various settings to analyze how different modeling and data decisions affect privacy.

Supernormal on the way to not-so-Normal Funding 🤑

Supernormal, a company offering technology that automatically transcribes and summarises meetings, has raised $10 million in funding led by Balderton, Acequia Capital, and byFounders VC - bringing their total funding to around $12.9 million. The new cash will further the mission of delivering end-to-end workflow solutions and developing next-generation tools based on foundational meeting data.

(Source: Techcrunch)

A Peek into the Future with Google AI 🫣

Google shares its progress and its vision for 2023 and beyond with respect to advancements in language, computer vision, multi-modal models, and generative machine learning models that will be applied to Google’s products to improve user experiences. Their overarching goal is to use these technologies to help people better understand the world around them and to make information universally accessible and helpful.

ChatGPT vs. Google vs. Baidu 🏇

Chinese search giant Baidu is set to launch a chatbot service similar to OpenAI's ChatGPT in March. The AI-driven tool, which does not yet have a name, will be embedded into Baidu's search engine and could become the most prominent entry in a race to adopt AI technology in China!

(Source: Bloomberg)

Tools of the Trade ⚒️

Bring 3D Designs to Life on your Fingertips 🤌

Here is a cutting-edge solution that revolutionizes the process of 3D prototyping. Mirage Canvas, an AI-powered 3D canvas, allows designers to craft stunning environments effortlessly. The platform features an AI-driven 3D search engine, offering users with a comprehensive and intuitive toolset for 3D design. Join the waitlist for early access to experience the ease of 3D prototyping like never before!

No more Lone Wolf Browsing 🌐

Don't navigate the vast internet on your own. Let Multi·ON be your companion and make the most out of your online journey. Available as a plug-in to web browsers, Multi·ON is the world's first AI web co-pilot powered by ChatGPT, making the online experience more efficient and convenient. You just need to instruct the chatbot what you want to do, and it’ll do the needful. (spooky!)

Build PPTs on-the-fly with AI 🛫

Prezo is a tool that uses AI to help you create stunning presentations 10x faster. With Prezo, you can:

• turn an article or memo into a slide deck,

• suggest variants with a customized tone for better storytelling, and

• generate captivating images to move people when you present.

This tool is designed to use AI for everything from creating slides and parts of a deck to generating visuals, charts, and themes, so you can tell your story in the best way possible.

Penning the Voice 🖊️

Turn your audio content into written form with VoicePen, an AI-powered tool for converting podcasts, webinars, or tutorials into blog posts. The process is simple: upload the audio file, make the payment, and the app will generate a blog post, transcription, and SRT file. VoicePen makes it easy to reach a wider audience and make your content more accessible.

AI-generated Seinfield that Just Doesn’t Stop! ⏱️

Nothing, Forever is a new AI-generated television show, streaming on Twitch, that started broadcasting in December 2022 and has been running ever since! The show features lo-poly versions of the classic Seinfeld characters, talking in robotic sentences. The artwork and laugh tracks are created by humans, while the dialogue, scene changes, and direction are all generated by AI algorithms, primarily using OpenAI's GPT-3.

Not sure if it can still be compared to human-produced Seinfield!

A Conversation with the Bible? 🤲

Imagine being able to have a conversation with the Bible! Well, now you can with BibleGPT. This innovative app allows you to describe your situation or ask a question in a simple text prompt, and you will receive specific verses from the Bible in response. Whether you need guidance, comfort or just want to explore the ancient text, BibleGPT is here to help. You can use it in English or Spanish.

Hot Takes 🔥

Python? Go? Wait, English is all you need! 🤔

After Wharton MBA, ChatGPT clears CFA too!! 😱

First Mover Advantage - Thing of past? 🐌

AI Meme of the Week 🤡

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!