Morning human brains, welcome back to your daily munch of AI news.

Here’s what’s up:

• AI that simulates drone footage 🏡 

• Stable Diffusion just got an update 🎨 

• AutoGPT will do your market research 🔍️ 

• GPT-5 will be released incrementally? 🦥 

Follow @BotEatBrain on Twitter for breaking news.

P.S. We are hiring researchers and writers, click here for more info.

APPETIZER

Simulating AI drone footage 🎥 

Last night, a group of AI researchers released Zip-NeRF, a research project based on NeRF tech (2D to 3D conversion AI) that borrows ideas from video rendering to train a neural model 22 times faster than previous versions.

Here’s what they cooked up:

This is not a recorded video.

To make this shot, the researchers walked around the place, took a bunch of photos (but not more than a couple per room), then fed them into the neural model which generated the 3D render of the place and the fly-through. Watch the full demo here.

The best part? NeRF uses regular pictures with no stabilized depth sensors or specialized equipment.

What this means: this is another example of the exponential progress curve in AI. NeRFs haven’t had their ChatGPT moment yet, but when they do, their applications for industries like AR and real estate will be unprecedented.

BUZZWORD OF THE DAY

FROM OUR PARTNERS

Scrape any site

Use WebScrape’s Chrome extension to scrape any website with just a sentence using natural language. Absolutely no coding required.

MAIN COURSE

Stability AI’s newest model 🖌️ 

Stability AI released its new Stable Diffusion XL model yesterday, promising photorealism and a significant jump in quality (play with it here).

Here’s what’s new:

1/ 2.3 billion parameters. That’s over a billion more than the previous version’s (Stable Diffusion V2.1) 900 million parameters.

2/ Consistently higher quality. You can generate high-quality images at 512x512 resolution using less detailed prompts now.

3/ Lower error rate. The model allegedly handles famously difficult parts of AI images like natural text and hands dramatically better than before.

4/ Editing interface. After creating an image, you can modify it with AI-based editing tools using the company’s new Dream Studio.

It’s not all rosy though. Having burnt through most of its cash fresh off last year’s $100M investment, Stability AI is balancing unimpressed investors and has begun approaching companies to integrate the new model into their existing products.

Our take: Stability AI’s future is uncertain, but this new model is a lot closer to Midjourney v5 in quality, and firmly puts the org back on the board.

MEMES FOR DESSERT

YOUR DAILY MUNCH

Think Pieces

The AI search business model might just destroy itself (1m).

GPT 5 will be released 'incrementally' - 5 points from Brockman statement [plus timelines & safety] (11m)

Startup News

Why Stability AI's XL Is a super-powered deepfake generator for businesses.

The AI startups to watch, according to European VCs.

The €20M Czech fund to elevate AI startup ecosystem.

Research

Optimizing linear correctors: improve the security of random number generators by using linear codes to process the raw bits.

APISENS: helping developers choose the best software artifacts by using the knowledge of people's emotions on social media.

Classifieds

Bye Blank Page: 10x your content writing with intelligent completions, suggestions, and text transformations (use BEB10 to get a 10% discount).

Tools

Sturrpy: build investor-ready financial models with AI.

MarkPrompt: generate GPT4 prompts for markdown documents.

AIHairStyles: discover how different hairstyles look on you.

Audio-bot: text to voice, in Latin and European languages.

TWEET OF THE DAY

GPT4 is now being optimized to prompt itself and conduct complex market studies from data points online.

Tag us on Twitter @BotEatBrain for a chance to be featured here tomorrow.

AI ART-SHOW

Until next time 🤖😋🧠