• NATURAL 20
  • Posts
  • OpenAI Q* might be REVOLUTIONARY AI TECH

OpenAI Q* might be REVOLUTIONARY AI TECH

How AI Is Tackling Complex Problems, from Protein Folding to Cryptographic Puzzles, and Its Potential for Future Breakthroughs

Today:

OpenAI Q* might be REVOLUTIONARY AI TECH | Biggest thing since Transformers | Researchers spooked

People used to say that neural nets couldn't do certain things, like solving complex math problems. But guess what? They're doing it now. There's this buzz about a new, super-advanced AI model from OpenAI called "Q star." People are going nuts trying to figure out what it's all about, and it's like the whole world's pitching in to solve this mystery.

Since GPT-4 hit the scene, there's been a lot of new ways to use these AI models, like making them think more like us humans when solving problems. Some smart cookies even got AIs to check each other's work. Orca 2 by Microsoft, showing that bigger, smarter AIs can teach the younger ones. The big buzz is about mixing the smarts of Google's AlphaGo with GPT-4 to make an even brainier AI.

Google’s new tools help discussion forums and social media platforms rank higher in search results

Google's rolling out new tools to help sites like social media and forums get more love on search results. You know how sometimes you search and get a bunch of useless, SEO-stuffed articles? Google's trying to cut through that mess. They started this thing called "Perspectives'' in May, which highlights real-deal posts from places like Reddit and Quora. Now, they're beefing it up so these genuine posts rank higher, making them easier to find.

Google's giving these websites a way to tell its search engine what's what, so their content shows up right and complete in search results. There's this new "ProfilePage" markup tool, so when creators post stuff, their profile info like name, photo, and how popular they are shows up in Google searches. This helps Google pick up the good stuff, even from smaller, less-known forums.

For website owners, Google's making life easier too. They're updating their Search Console with new reports to track any issues or successes with their site's content. Plus, there's this Rich Results Test for them to check if their markup is on point.

The big picture? Google's trying to stay ahead of the game. With AI cranking out tons of fake or low-quality content, Google's worried its search results might get swamped with junk. There's chatter about AI being misused for stuff like stealing web traffic, and Google wants to keep its results legit and useful. They're also dabbling in AI themselves with a new answer engine and experimenting with letting users add notes to web pages. It's like they're trying to build their own Reddit right in Google. But for now, it's just a test.

Salesforce and AWS Expand Partnership for Customers to More Easily Build Trusted AI Apps, Deliver Seamless CRM Experiences, and Bring Salesforce Products to AWS Marketplace

Salesforce and AWS, two big tech players, are joining forces in a major way. They're making it a breeze for folks to build AI apps that are on the level and give customers a smooth experience with their CRM software. Plus, Salesforce is going to start selling its stuff on AWS's online shop.

For the first time, Salesforce will offer its products, like Data Cloud and Service Cloud, on the AWS Marketplace. This means AWS customers can easily get their hands on Salesforce products. They're also making it easier to handle data, with less need for customers to mess around with techy stuff to connect Data Cloud and AWS data services.

They're also ramping up customer service, making it smoother for people to get help through Salesforce and AWS. And for the tech heads, they're beefing up the tools available for building AI-first apps.

Some big-name customers like State Farm and 3M are already seeing the benefits, saying this partnership is helping them serve their customers better.

Sports Illustrated deletes articles published under fake author names and AI-generated profile photos

Sports Illustrated, a big-time sports magazine, got caught with their pants down. They were using fake names and computer-made faces for some of their articles. This sketchy stuff got spotted by Futurism, another publication. Turns out, these so-called authors were ghosts; nobody could find them anywhere except on Sports Illustrated's own website. Even weirder, their faces were just AI creations you could buy online.

When Futurism started poking around, some of these mystery writers just up and disappeared from Sports Illustrated's site. Then, their articles popped up under different names, but with the same deal - fake people, AI faces.

The Arena Group, the big boss behind Sports Illustrated since 2019, said these shady articles were from AdVon Commerce, a third-party company. They said AdVon had writers use fake names to keep their privacy, which Sports Illustrated wasn't cool with. They dumped the content and kicked AdVon to the curb.

AWS’ transcription platform is now powered by generative AI

AWS just beefed up their Amazon Transcribe thing big time. Now it's got this generative AI magic that can handle 100 different languages. They dropped this news at their AWS re:Invent shindig.

 They trained this tech on a ton of audio from all these languages. They made sure not to play favorites with the popular languages, so even the less common ones get the same level of love and accuracy.

Last year, they could only do 79 languages, but now they're saying they're hitting between 20 to 50 percent accuracy, depending on the language. Plus, they've got some neat features like auto punctuation, recognizing speech in noisy spots, and even understanding speech in videos.

This tech isn't just for transcribing stuff. It's also making waves in their Call Analytics platform. This is used by call centers to get the lowdown on chats between customers and agents. Now, with the AI upgrade, it sums up these chats so managers don't have to slog through the whole convo.

Cradle’s AI-powered protein programming platform levels up with $24M in new funding

Cradle, a biotech and AI startup have figured out a way to use AI to mess around with proteins. They've compared proteins to some sort of alien code, and their AI is cracking it. Basically, they're teaching computers to understand and mess with these proteins, which is something humans can't easily do.

They're not just talking big; they've got some major players like Johnson & Johnson paying attention. Making new proteins usually takes a boatload of time and lab experiments, but Cradle claims they can speed things up big time. They didn't give all the deets but showed off by making a bunch of variants of this enzyme, T7 RNA polymerase, that can handle the heat better. Usually, you'd get a tiny percent of these tweaked molecules working right, but Cradle's hitting like 70% success. That's like running four or five experiments at once!

They're working on a bunch of other stuff like soil decontamination, stuff for cultured meat, and even treatments for diseases. The big deal here is their AI models are doing better than traditional methods. This isn't just for medicine either – it's got potential in food and industrial areas too. And the best part? You don't need to be some computer whiz to use their system; scientists and labs can get right on it.

XTX Markets launching $10 million 'Artificial Intelligence Mathematical Olympiad Prize' 

XTX Markets is shaking things up with a big-time challenge. They've set up this thing called the "AI-MO Prize," where they're throwing down $10 million to encourage folks to build an AI that can ace the International Mathematical Olympiad (IMO). The grand prize? A cool $5 million for the first AI that can score a gold medal in an AI-MO approved competition.

They're also dishing out $5 million in smaller prizes for AIs that hit certain milestones on the way to the big win. They're setting up an advisory crew with math, AI, and Olympiad pros to make the rules. The first competition is set for early 2024, with a progress report at the 65th IMO in England.

PhysicsX emerges from stealth with $32M for AI to power engineering simulations

there's this new AI company called PhysicsX, started by a couple of brainy physics types, including this Formula One engineering whiz. They're stepping out with a big focus on using AI for real-world stuff, like building and managing stuff in big businesses.

PhysicsX is based in London and they've cooked up an AI system that helps engineers run simulations. This is a big deal in fields like cars, flying stuff, and making materials, where figuring things out before you build them can really slow things down. They're launching with a hefty $32 million in their pockets.

The cash comes from a Series A funding round led by General Catalyst, with a bunch of other heavy hitters pitching in, like Standard Industries, NGP Energy, and even a big name from KKR, Henry Kravis. They're planning to use this dough for growing their business and making their AI even better. It's the first time they've taken outside money.

AI and robotics company BRAIIN acquires India’s Exato.ai

BRAIIN, a big shot AI and robotics company from the US, has stepped into India by snapping up Exato.ai. This Exato outfit is all about making customer experiences smooth and digital. BRAIIN's looking to set up shop in India using Exato as its springboard.

The big boss at BRAIIN, Natraj Balasubramanian, is stoked about moving into India. He's talking about dropping a cool $100 million in the first eight months alone. He's got high hopes for India, seeing it as a gold mine for tech growth, especially in stuff like data crunching, smart devices, and more AI wizardry.

BRAIIN's got its fingers in a few pies already, helping businesses in telecom, farming, and service industries get smarter with AI, the Internet of Things, and robots. They're planning to buy up more companies to boost their tech game and are even chatting with the Indian government about bringing their tech to big-scale farming. They're eyeing the tea and rubber plantation game too.

AI under the microscope: the algorithms powering the search for cells

In the world of biology, folks are relying on computers to spot cells, and it ain't as easy as it sounds. Scientists have been training these machines to recognize cells in images, but it's been a tough road. They've had to customize algorithms for each experiment, which is a real pain.

But here's the good news: thanks to deep learning, things are lookin' up. These algorithms, like the U-Net, are making the process way faster. They're getting better at recognizing different types of cells and even figuring out their shapes.

Nowadays, instead of hand-holding, scientists use machine learning tools like CellProfiler and ilastik. They show these tools examples, and the software learns from 'em. This makes the whole process more user-friendly.

The real trick here is the training. The better the data and labels, the better the results. Some smart cookies are even using a "human in the loop" approach, where algorithms do most of the work, and humans just check their work. It's a real time-saver.

AI may spare breast cancer patients unnecessary treatments

Northwestern Medicine's got this new AI that could really change the game for folks dealing with breast cancer. This smart tech is way better at figuring out how the cancer's gonna play out than even the top docs. It zeroes in on both the bad (cancer) and the good (non-cancer) cells in the tissue, something that hasn't been done before.

Why's this a big deal? Well, it means some patients, who the docs thought were in deep trouble, might actually have a fighting chance without going through the wringer with heavy chemo. This is huge 'cause chemo's no walk in the park – it can knock you out with side effects like nausea or, worse, heart damage.

Normally, pathologists eyeball the cancer cells to decide on treatment, but this AI's showing that the other cells hanging around the cancer are key to predicting what's gonna happen. The AI's not just smart; it's clear too. It gives the pathologists a heads-up on what it's thinking, breaking down the patient's prognosis based on different cell types. This could lead to more tailored treatments for patients.

They trained this brainy AI with a boatload of data from human experts and are now testing it out to make sure it's ready for prime time. Plus, they're cooking up versions for different types of breast cancer. The goal? Better predictions, better understanding of the cancer, and hopefully, a smoother ride for patients, especially those in community hospitals who might not have access to top-tier cancer experts.

5 ways Amazon is using AI to improve your holiday shopping and deliver your package faster

Amazon is using some smart tech to make your holiday shopping a breeze and get those packages to you lightning fast:

  1. Stocking Smarts: Before you even hit 'buy', Amazon's using this brainy system called SCOT. It's like a crystal ball for shopping, predicting what folks will want to buy. It's super sharp at figuring out what stuff to stock and where, so you get your goodies quicker.

  2. Robot Helpers: Amazon's got this army of smart robots in their warehouses. These bots know how to pick out and check your stuff before it's packed up. They even train with make-believe scenarios to get better at handling the holiday rush.

  3. Speedy Sorting: Once your order's packed, a robot named Robin steps in. It's got eagle eyes for sorting different packages, deciding the best way to grab 'em without making a mess.

  4. Road Surprises: When your package hits the road, things can get unpredictable, like early trucks or bad weather. But Amazon's AI is like a weatherman mixed with a fortune-teller, helping them stay one step ahead.

  5. Finding the Best Path to Your Door: Planning the delivery routes is a real brain-teaser, like a giant puzzle. Amazon's got a bunch of smart systems working together to figure out the fastest way to get your package to you.

So, basically, Amazon's using some really smart tech to make sure your holiday shopping is as smooth as butter and your packages arrive faster than ever. Cool, huh?

What'd you think of today's edition?

Login or Subscribe to participate in polls.

What are MOST interested in learning about AI?

What stories or resources will be most interesting for you to hear about?

Login or Subscribe to participate in polls.

Reply

or to participate.