- NATURAL 20
- Posts
- Meta & Hugging Face's Startup Program
Meta & Hugging Face's Startup Program
Meta and Hugging Face are scouting for startups ready to disrupt the AI scene
Wes Roth
November 09, 2023
Today:
Meta taps Hugging Face for startup accelerator to spur adoption of open source AI models
Meta's basically joining forces with Hugging Face and Scaleway over in Paris to kick off a new program for AI startups. They're aiming to shake things up by encouraging a more open, together-we-can style of creating AI in France. The timing's pretty interesting, given all the buzz about needing rules for AI and the tug-of-war between keeping AI tech under wraps or sharing it with everyone.
Big tech firms like Microsoft and Google are shouting for more rules, talking about how AI could go rogue if it's just out there for anyone to mess with. But on the flip side, there's this idea that these companies might just be playing the fear card to get regulators on their side, keeping their tech secrets safe from the competition.
Meta's not buying it. They want to keep AI open and shared. Their top AI guru, Yann LeCun, is one of many big brains saying, "Hey, let's not rush into rules that could backfire and stifle the cool stuff we could all come up with."
Meta's got some serious street cred in the open AI scene and has been sharing their own AI creations. Hugging Face, on the other hand, is like this giant in the open source AI playground, offering up their own takes on stuff like ChatGPT.
They’re launching this “AI Startup Program” where companies can throw their hats in the ring until December 1. The winners get to be part of this cool accelerator program next year, working with Meta and Hugging Face to make AI that’s open for all to play with.
GitHub teases Copilot enterprise plan that lets companies customize for their codebase
GitHub's got this new trick up its sleeve for the big leagues, rolling out a fancy version of its Copilot tool that's gonna let companies make it their own. At their big dev shindig, GitHub Universe, they spilled the beans on this and some other cool stuff, like a partner program and the lowdown on when their Copilot Chat will be ready to chat with everyone.
Copilot Chat, for those out of the loop, is like a buddy in your coding toolbox that helps you figure out what's going wrong in your code and offers up fixes and feedback right there while you're working on it. It's smart, too—uses OpenAI's GPT-4 to get the job done. It's been around in beta for a bit, and they're looking to make it available for everyone in December, but they're keeping the exact date under wraps.
Well, for $39 a month starting February 2024, companies can customize this Copilot Chat to know their code like the back of its hand, give suggestions on the company's private code, and even chip in on stuff like pull requests right on GitHub's website.
On top of that, GitHub's planning to bring Copilot Chat to its mobile app and JetBrain's IDEs, 'cause people really wanted it. They're also starting up this Copilot Partner Program to hook up with other toolmakers and create plug-ins that'll make Copilot even better.
Last but not least, GitHub's teasing this thing called Copilot Workspace, which they say will help turn ideas into running code in no time. It's coming out sometime in 2024.
AI Startup Anthropic to Use Google Chips in Expanded Partnership
Google's just beefed up its partnership with this hotshot AI startup called Anthropic. They're the new kid on the block, challenging the big dog, ChatGPT. Google’s throwing a cool $2 billion into the ring for them.
Now, Anthropic's got this brainy AI named Claude, and they're hitching it to Google’s latest and greatest TPU v5e chips. These chips are no joke—they're like the fifth-gen of Google’s homegrown tech and pack a serious punch, more than twice as fast as the old ones.
But Claude’s got a big appetite, it needs a truckload of processors to get going. Google’s got this nifty thing called Multislice that lets Claude stretch its legs over tens of thousands of chips. They say it’s gonna make AI safer and more secure for everyone.
The word on the street is that Anthropic’s planning to drop over $3 billion on Google Cloud goodies over the next few years. And they ain’t just buying chips—they’re looking at other gear too, plus some cyber-safety tools to keep the hackers at bay.
Anthropic isn’t new to Google Cloud—they’ve been buddies since 2021, running one of the biggest Kubernetes setups you’ve ever seen, and crunching numbers with Google’s data tools.
Google is bringing its AI-powered search to more than 120 new countries and territories
Google's got this new thing, the Search Generative Experience, and they're rolling it out big time—over 120 new places are getting it, like Mexico, Brazil, and South Korea. They're also throwing in support for four more languages, which is pretty cool. If you wanna give it a whirl, you gotta opt into their Search Labs thing and use Chrome or the Google app.
The Google folks are keeping mum about whether it'll work on other browsers down the line, but they're hinting there's more to come. There's talk that this feature might wrap up in December 2023, but the Google team is just looking for feedback for now and not spilling the beans on what's next. But the way the big boss at Google talks about it, it sounds like this feature is here to stay.
Plus, Google's tossing in some handy-dandy extras, like a "ask a follow up" box right in the search results, and helping you figure out words with more than one meaning when you're looking for a translation. They're also making it easier to get the lowdown on techy and health terms just by hovering over them. This is all starting to pop up in the US first.
Amazon Is Reportedly Racing To Build An AI Model Called Olympus To Take On ChatGPT And Bard
Amazon's stepping up its game in the tech showdown, gunning to outdo the likes of ChatGPT and Bard with a fresh AI project they're calling "Olympus." This new kid on the block is said to be a heavyweight, packing twice the punch of GPT-4 with its two trillion parameters. They're not just throwing their hat in the ring; they're laying down some serious cash, with a $4 billion shout out to AI startup Anthropic.
Word on the street is, Amazon's looking to boost its online shopping experience and smart speaker smarts by baking "Olympus" right in. There's buzz that we might catch a glimpse of this beast by December. Amazon's boss, Andy Jassy, is keeping a close watch on the crew crafting these mega AI models.
Not ones to sit on the sidelines, Amazon's been hustling to bridge the gap with "Project Nile," aiming to make shopping chitchat a breeze, and souping up Alexa to be more of a gabfest guru and smarty-pants helper.
While Alexa's new tricks are still under wraps with just a few getting a sneak peek, Amazon's hinting that they might start charging for the VIP AI treatment. They're playing it coy with the details, but it's clear they're not just in it to play — they're in it to win.
Facebook and Instagram will require political advertisers to disclose AI deepfakes
Meta, which is the big boss of Facebook and Instagram, is layin' down some new rules for the political ad game. They're sayin', starting next year, if you're putting out a political ad, you gotta be upfront if you've used any fancy computer tricks—y'know, like AI that makes stuff up or changes what really went down. They're doing this to keep things real, especially with the big election coming up in 2024.
Now, if you're just tweaking a photo a bit, like fixing the lighting or cropping it, they're not gonna hassle you. But the big deal here is that Meta is trying to keep the digital playground fair by not letting their own AI toys be used for political ads.
Samsung’s Galaxy S24 will likely include on-device generative AI called Samsung Gauss
Samsung's cookin' up a fresh new AI called Samsung Gauss that's gonna be part of the Galaxy S24 phone lineup, set to drop in early 2024. This smart cookie of an AI can do a bunch of cool stuff like whipping up images, helping you write emails, and even getting down and dirty with some code. The big deal? Some of this magic can happen right on your phone, no need to bug the cloud for help.
This Samsung Gauss isn't just some random name; it's a shoutout to a math whiz from way back, Carl Friedrich Gauss. They're rolling out different flavors of this AI for language, coding, and images.
Now, Samsung's been hinting that they're gonna sprinkle some of this AI goodness into the main features of their phones starting with the S24. This might just give the new Galaxy an edge, 'cause it'll handle some image and language tasks on its own, without leaning on cloud giants like OpenAI.
But hold up, Samsung's also jazzing up their Bixby assistant with some of this AI talent for their home gear, though it's gonna be more low-key than the heavy-hitting ChatGPT. They're aiming for a chiller vibe, where Bixby catches your drift even when you go off-script.
Fakespot Chat, Mozilla’s first LLM, lets online shoppers research products via an AI chatbot
Mozilla scooped up a company called Fakespot to help shoppers online tell if reviews are the real deal or just a bunch of baloney. They're launching a chatbot—think of it as your virtual buddy that's got the inside scoop on products. It's a bit funny, 'cause it's using AI to fight off fake reviews, which are sometimes whipped up by similar tech. The bigwigs have been cracking down on these phony write-ups, with TripAdvisor scraping off over 20,000 suspicious reviews.
Fakespot's not new to the game; they've been using AI to give shoppers the straight talk on what's good and bad about a product without wading through a sea of reviews. Now, they're testing a chat feature where you can chat up an AI about stuff you wanna buy, just like you'd yak with a store clerk. It digs through reviews to fish out the fakes and doesn't need your life story to help you out.
You can check this out either through their Analyzer or a browser extension if you're looking at stuff on Amazon. It's all about helping you make smart buying choices without spending an age researching. Mozilla's also throwing cash at other AI projects that are on the up-and-up, aiming for AI that's on the level.
Alibaba CEO Eddie Wu reaffirms AI commitment, aims to turn e-commerce empire into 'open tech platform'
Eddie Wu, the new big boss at Alibaba, is shaking things up. He's steering the ship towards AI big time, saying it's not just about online shopping anymore. They're opening up their tech to everyone, aiming to give a leg up to all sorts of businesses with their AI tools.
Wu's betting that AI is the next big thing for China's economy and he wants Alibaba at the heart of it, helping the little guy and big players alike to get smarter and work faster. He's already got Alibaba Cloud rolling out its own version of ChatGPT in China, where the American stuff isn't on the menu. Wu's vision is a future where AI's in everything, making life and work a breeze.
Hugging Face has a two-person team developing ChatGPT-like AI models
Hugging Face, the AI shop that's all about data crunching and sharing AI smarts, has this tiny but mighty team called H4. They've got this simple game plan: take what made ChatGPT a hit and make it something anyone can mess around with, using stuff that's out there for grabs. The H4 duo kicked things off after ChatGPT made waves, thinking about how to cook up something just as slick without keeping it under wraps.
Their big brain work involves training AI to play nice and act smart based on what people tell it, or even other AIs. They're the whizzes behind some nifty AI models that anyone can tinker with. To make their magic happen, they tap into a serious pile of Nvidia's fancy computing muscle.
These two, Lewis and Ed, work out of Europe but have a bunch of Hugging Face pals to help out. They like being a small outfit because it lets them move fast and stay on top of things. They're not alone though; they team up with other brainy groups for some projects.
Lately, they've been digging into ways to make sure AI behaves like we want it to and sharing their tools to see if they're really up to snuff. They even dropped a guidebook with all their secrets for their latest AI, Zephyr, and plan to keep doing that with their new stuff.
Now, Hugging Face has some big-time backers with deep pockets, but H4 isn't about making a quick buck. Their gadgets help out with Hugging Face's big-kid program that helps businesses build their own AI toys.
Scale Launches New AI Safety Lab, Led By Former Google Bard Researcher
Scale AI's on the move! They're kicking off a new AI safety lab, SEAL, and it's all about keeping AI on the straight and narrow. Think of it like a test track for AI to make sure it doesn't go rogue. They're bringing in Summer Yue, the whiz from Google's chatbot team, to run the show. SEAL will whip up tools for companies to spot any slip-ups in their AI before it hits the streets.
Scale ain't new to the game. They started by helping cars drive themselves and now they're into all sorts of stuff, from shopping to national defense. And their CEO, young gun Alexandr Wang, is about to join the Forbes crew. He's all about understanding AI's weak spots to make it better and safer. With everything moving at lightning speed, he's saying we gotta stay on our toes and SEAL's here to help with just that.
Liquid biopsies and AI give insight into eye health and aging
Scientists have figured out a new way to peek into our peepers to check on our health and how fast we're aging. Instead of cutting out bits of our body to study, they're looking at the goop (eye fluids) that's already there, especially during eye surgeries. By checking out the proteins in this goop, they can tell if something's going wrong with our eyes, like if our blood vessels are getting messed up by diabetes or if we're getting eye inflammation.
They tried this out on 120 folks getting eye surgery, some with eye problems and some without, and found a bunch of proteins that could be like red flags for diseases. They even saw signs that might hint at Parkinson's disease way before it's usually caught. Plus, they're using computer smarts to guess how old our eyes are, showing that eyes with certain diseases might look way older than they should.
The cool part is this trick might work for other parts of the body that have their own fluids, like our brains, kidneys, lungs, and joints. It's like having a crystal ball that lets us see into our bodies without having to poke around too much. And the cash for this comes from some big-name health and research groups that are all about keeping our eyes and the rest of us in tip-top shape.
AI could steal over 80 million jobs in the next 5 years, but these experts share how to adapt before it's too late
Robots and AI could snatch up like 80 million gigs in the next five years. That's a heap of jobs, right? But hold up, there's hope. Some brainy folks got tips on how to stay in the game. They're saying to buff up on the human touch—stuff computers can't fake. Things like being a good talker, a team player, and a boss who knows how to get folks to play nice together.
Now, these AI bots, they're smart, but they ain't got nothing on a well-rounded human. You gotta get curious about stuff outside your usual turf. Read up on things nobody else is checking out. It might feel weird, but that's the point.
And here's the kicker: you gotta get slick at telling these AI what to do. You know, typing in those prompts that make them spit out the good stuff. One pro said she did this and bam, her cash flow jumped big time. So, learn to ask the right way, and you could be golden.
It’s Never Been This Easy To Build A $1 Billion AI Comapny
Building a big-time AI biz worth a cool billion isn't just for the heavy hitters anymore. Folks like Sam Altman at OpenAI got the ball rolling with big bucks from Microsoft and kept their tricks under wraps. But now, Kai-Fu Lee's new outfit and a fresh company called Mistral AI are changing the game by sharing their code with everyone.
Lee's startup hit a billion-dollar mark in under eight months, and it's all about being open. They let the world see how their AI, named Yi-34B, learns and stacks up against others. And guess what? It's actually sharper than some of the big dogs' tech.
Meanwhile, Mistral AI from Paris is also all about that open-source life. They're so into it, they're giving away their AI smarts for nada. And they're talking big money, like $400 million big, to become Europe's answer to OpenAI.
Turns out, you don't need a mountain of cash or secret sauce to get ahead in the AI world. Playing it open and honest with your tech can get you there just fine. The smart folks at Hugging Face, a platform where you can show off your AI, are all about this approach.
What'd you think of today's edition? |
What are MOST interested in learning about AI?What stories or resources will be most interesting for you to hear about? |
Reply