- Writing Inc
- Posts
- š¤ Midjourney censors AI-generated photos of Xi Jinping - hereās why
š¤ Midjourney censors AI-generated photos of Xi Jinping - hereās why
PLUS: Kitchens swap chefs for AI
Would you go to a restaurant whose head chef is an AI?
Second question: would you tip?
š¤ Top Stories
𤯠A restaurant without any chefs, thanks to AI

Source: Eater
Picture a restaurant with no chefs.
CloudChef is a Silicon Valley startup that does exactly that. Instead of chefs, they staff their kitchens with workers who don't know anything about food.
We hear you. That's a departure from the last...6000+ years. So, how does it work?
First, they record a real chef preparing a dish. They use video cameras, as well as infrared cameras and scales.
After that, they translate all that info into a set of machine-readable instructions.
So let's say it's Friday night and we've got an order for that meal.
The software walks them through weighing ingredients, sauteing vegetables, and searing steaks. One worker covers prep, the next covers the actual cook. Neither of them know anything about cooking. The AI guides them through it all.
Sound dystopian?
It is. They call themselves the Spotify of food, partly because chefs get a royalty of any dish made using their recipe.
Sounds nice for everyone, except...the workers they're replacing.
Such is AI.
š¬ Midjourney bans photos of Xi Jinping
It seems not all political satire is created equal.
Midjourneyās latest update banned any images that involve the president of China.
Lame.
Sure, this is in the wake of fake Trump and fake Pope images sprouting up. But this goes a step further. By singling out the leader of China, many argue this amounts to censorship.
The CEO disagrees.

Source: Discord via Decrypt
His defense is that heās actually protecting the (fair amount of) Chinese users on their platform. Political satire isnāt looked upon as kindly in China.
āRandom people on here doing Chinese political satire does very little to add to anythingāthe cost-benefit-analysis seems clear.ā
They call it āminimizing dramaā.
Others call it caving in.
š Meta AI finally has something neat for us
Today we're releasing the Segment Anything Model (SAM) ā a step toward the first foundation model for image segmentation.
SAM is capable of one-click segmentation of any object from any photo or video + zero-shot transfer to other segmentation tasks ā”ļø bit.ly/433YuBI
ā Meta AI (@MetaAI)
1:01 PM ⢠Apr 5, 2023
Round of applause.
Their new tool called Segment Anything is scary good at one thing and one thing only.
Taking any image and cutting out portions of it.
They have a demo here. Upload a busy picture - like a packed Times Square - and watch as it selects individual people with ease.
The best part? Itās open source. Actually open source. Not fake open source like their large-language model.
Meta has a long way to go to catch up to Google and Microsoft. Projects like this might help.
š AI tools, tips, links
ChatGPT for every website. Learn more
Voice samples generated by AI. Spooky stuff. Learn more
Google AI team member says he built Metaās SAM first⦠Learn more
Testing Midjourneyās permutation feature. Learn more
Another AI travel planner! Learn more
Close enough, ChatGPT.

Source: Reddit
š¤ Got feedback for us?
Head over to our Twitter, @RobotRockDaily, and let us know what youād like to see more or less of.