OpenAI's Sora, Gemini's Update, Stable Diffusion's Successor, and MORE!

PLUS HOT AI Tools & Tutorials

Feb 16, 2024

👋 Hey, I’m Stepan and welcome to a ✨ news edition ✨ of Creators’ AI. By subscribing, you directly support Creators' AI's mission to deliver top AI insights & practical knowledge without ads or clutter. Your subscription allows us to grow our dedicated team and curate the most important AI Tools, Stories, and Tutorials in one place. - Stepan

Welcome to our weekly newsletter! For the last few weeks, Google, with its ImageFX & MusicFX, Gemini, and other products, has been the primary trend in AI news. Yesterday, it seemed that the company would continue its march with the Gemini 1.5 update. Still, suddenly, Sam Altman and OpenAI burst onto the scene with Sora, a huge text-to-video model.

And it's no exaggeration to say that this AI has blown up social media.

Today, we'll see what Sora can do and what kind of videos it allows us to create. In doing so, we won't forget the other significant news. And, of course, you'll get some AI Tools, guides, and meme & tweet of the week. Let's get started.

This Creators’ AI Edition:

Featured Materials 🎟️
News of the week 🌍
Useful tools ⚒️
Weekly Guides 📕
AI Meme of the Week 🤡
AI Tweet of the Week 🐦
(Bonus) Materials 🎁

Share this post with friends, especially those interested in AI stories!

Featured Material 🎟️

Sora was announced this Thursday on Sam Altman's X-account and the OpenAI website. It's a text-to-video model that uses short text descriptions to create realistic videos. I mean, it's very realistic.

You'll see examples below, but first, let's talk about Sora itself.

According to OpenAI, Sora is a model that can generate videos up to a minute long while maintaining visual quality and adherence to the user's prompt. It can create complex scenes with multiple characters, specific types of motion, and precise object and background details (and it's true!). The most important feature is that the model understands not only prompts but also how these objects exist in the physical world.

We’re teaching AI to understand and simulate the physical world in motion, with the goal of training models that help people solve problems that require real-world interaction.

At the same time, OpenAI recognizes that the model has weaknesses. Sora may have trouble accurately modeling the physics of a complex scene, and it may not understand specific causes and effects. For example, the company cited a situation where a person in a video bites a cookie, but the cookie remains intact afterward.

To celebrate the launch of Sora, Sam Altman ran an unusual promotion at X. He invited users to send him captions that he would use to generate videos. The OpenAI CEO told enthusiasts not to hold back and come up with really complex descriptions.

Let's take a look at what they got:

Caption:

A wizard wearing a pointed hat and a blue robe with white stars casting a spell that shoots lightning from his hand and holding an old tome in his other hand

Result:

Caption:

A instructional cooking session for homemade gnocchi hosted by a grandmother social media influencer set in a rustic Tuscan country kitchen with cinematic lighting

Result:

You can see more results from Sam Altman here. For now, look at two more videos generated by Sora.

An important addition that Altman didn't mention is that Sora does a great job of generating the physical as well as the virtual world. Just take a look at how this model handles with Minecraft’s lets play:

The results are really impressive (And, honestly, a bit scary). To be a little more pragmatic, let's review the small details. Here, for example, is an approximated snippet of a video of a girl walking through a busy city.

Take a look at the movement of her legs:

Yes, the legs are tangled in places, which is confusing. You can see a similar effect in the cat video here. Look at the paws. As you can see, OpenAI still has work to do.

Sora is being distributed to a limited number of creators. When the model will be publicly available is still unknown. We'll be waiting and keep you posted!

Well, we still have time to believe social media content (just don't believe it too much!). The question is, for how long?

My bet: we're one year away from the point where videos stop being proof of anything. In the meantime, new models will open the door to create and consume endless variations of content.

News Of The Week 🌍

Google unveiled next-gen AI model Gemini 1.5

Sorry, Google, but you picked the wrong week. Two months after the launch of Gemini, the company has announced a new version. It is a more robust model on the same level as Gemini Ultra. It is already available to developers and enterprise users. Consumers, on the other hand, will have to wait.

Google promises a rollout in the coming months.

The German nonprofit company develops public voice assistant

Large-scale Artificial Intelligence Open Network (LAION), a German nonprofit organization, has announced a new project. The company wants to create a "fully open" consumer-oriented voice assistant. The authors say there is still no assistant with an extensible enough architecture to take full advantage of the new GenAI technologies. And LAION is poised to solve that problem.

Stability AI reveals a more powerful model, Stable Cascade

The leading open-source image generation model, Stable Diffusion, has a successor. It's called Stable Cascade and claims to be a more powerful and faster model. The model allows you to modify already created images or increase the resolution of existing pictures.

Unlike Diffusion, Cascade uses three different language models at once.

OpenAI is working on a Web Search Product to compete with Google

According to The Information, ChatGPT developers are developing a new search engine. Insiders say the new product may be partially powered by Microsoft's search engine, Bing. Whether it will be part of ChatGPT or become a standalone system is still unclear.

But the news is important: OpenAI is one of the few companies able to impose tangible competition on Google.

Love is deceptive: 10 out of 11 romance chatbots are selling your data

Researchers at Mozilla *Privacy Not Included have figured out that AI girlfriends and boyfriends are storing personal information about their users and then selling or putting that data in the public domain. Mozilla studied 11 romantic AI chatbots, including popular apps like Replika, Chai, Romantic AI, EVA AI Chat Bot & Soulmate, and CrushOn.AI. And each of them has been labeled "Privacy not enabled".

These are the worst chatbots that Mozilla has ever tested.

Airbnb has started using AI to create the "ultimate concierge"

Last November, Airbnb acquired GamePlanner.AI, a company created by Siri's co-founder. As TechCrunch writes, the deal's value amounted to $200 million. Then Airbnb did not disclose the purpose of the purchase, but now the reason has become known. According to CEO Brian Chesky, his company will use GamePlanner.AI technology to create one of the "most innovative AI interfaces."

This model will help Airbnb improve the customer experience.

Subscribe to stay in tune with AI breakthroughs!

Useful Tools ⚒️

Goody-2 – The world's most responsible AI model with ethical principles.

Meals.chat – An AI assistant for tracking diet quality.

Squad AI – The product strategy tool for everyone.

RDMC AI – Your own AI digital marketing agency.

Lindy AI – Build your own AI agents in 5 minutes with no code.

Lindy AI allows you to create an AI employee to suit any taste. The platform generates recruiting, sales, consulting, and more. With Lindy, you can automate communication with customers and suppliers, and forget about filling out and checking documents. It integrates with third-party systems.

We’ve also made a list of the best GPTs for Startups & Marketing:
23 GPTs For Marketing & Startups : GPT Store Selection
Creators' AI
·
Jan 23
Read full story

Creators’ AI could be a valuable gift for your friend, colleague, or family member. Gifting books is bright, but giving an AI newsletter is a super move 😎

Give a gift subscription