Microsoft's $5 Billion AI Investment

Today:

Microsoft announces A$5 billion investment in computing capacity and capability to help Australia seize the AI era

Microsoft’s throwing down a cool A$5 billion in Australia, aiming to be the big kahuna in the AI game down under. They’re bulking up their cloud and AI muscles, turning 20 datacenters into 29 across Canberra, Melbourne, and Sydney. Not just a pretty penny, it's their fattest cheque in 40 years in Australia.

They’re not just flashing cash; they’re opening a Datacentre Academy with TAFE NSW, gunning to smarten up 300,000 Aussies in the cloud and AI hustle. On the cybersecurity front, they’re joining forces with the Australian Signals Directorate, cooking up some digital armor to keep the bad guys at bay.

This shindig lines up with the Aussie Prime Minister's U.S. tour, rubbing elbows with Microsoft’s top dogs. The plan? Crank up Australia's computing power by 250% in two years, cashing in on the cloud boom and riding the AI wave to potentially add A$115 billion a year to the economy by 2030.

READ THE ARTICLE ON MICROSOFT

Qualcomm’s next big Snapdragon chip has leaked, and it’s full of AI features

Looks like Qualcomm's got a new Snapdragon chip on the horizon, and it's playing its cards close to the vest... but not close enough. Details are spilling out early, and they're all about AI. The Snapdragon 8 Gen 3 is promising to bring some nifty AI tricks to your next Android phone, like playing magician with your videos and photos, and cranking out fake backgrounds like nobody's business.

This chip is flexing with its upgraded Hexagon neural processor, claiming to be nearly double the speed of its last act. Last time we checked, Qualcomm was already showing off with its Snapdragon 8 Gen 2, running fancy AI models like a champ. But now? It's all about AI, all the time.

READ THE ARTICLE ON THE VERGE

Australia's Canva takes deep dive into AI to help regain valuation

Canva, the Aussie design software hotshot, is throwing its hat in the AI ring, hoping to jazz up its toolkit and charm big-time clients. After taking a financial nosedive, they're banking on these smart tools to stay in the startup major league. Co-founder Cameron Adams is talking big, saying AI is a game changer, letting folks whip up designs on the fly without breaking the bank.

They've rolled out a bunch of AI tricks, like turning rambles into neat summaries and cooking up photos and videos from thin air. It's a bold move, especially when they're duking it out with the likes of Adobe and some fresh-faced startups. Remember, Adobe's on the move too, having tried to scoop up Figma for a cool $20 billion.

READ THE ARTICLE ON NIKKEI

Anthropic tests AI rules for the people, by the people

Amazon's pet project, Anthropic, known for their Claude chatbot, is trying to jazz things up. They're creating a rulebook for AI based on what John and Jane Doe think. Basically, a "people's constitution" for AI. Why? Because it's a hot mess when AI starts thinking it knows better than us, especially with all the biases it can pick up.

Here's the tea: Anthropic did a poll with 1,000 regular folks. They got 75 rules from this. Now, Anthropic already had their own 58 rules for Claude. Guess what? Only half of those matched the people's rules. So yeah, there's a gap between Silicon Valley's big brains and regular Joe's coffee chat.

They tested both sets of rules on a baby version of Claude. Both did alright - around 85% accuracy. But the people's version was less likely to make a fool of itself by being biased on things like age or religion.

READ THE ARTICLE ON AXIOS

Microsoft opens early access to AI assistant for infosec, Security Copilot

Microsoft's letting folks peek behind the curtain at their new cybersecurity whiz kid, Security Copilot. This tool's been in the wings since March, promising to turn security drudgery into a breeze, freeing up a whopping 40% of time spent on routine tasks. It's like having an extra pair of hands on deck, especially for the greenhorns in the security biz.

The spiel is that even if your team isn't the '97 Bulls of cybersecurity, Security Copilot will make them look good, providing a step-by-step game plan for tackling incidents. Critics are itching to see if it lives up to the hype.

Copilot's got a few tricks up its sleeve, like whipping up reports in plain English and breaking down threats to see how they tick. It’s even got a knack for crafting queries in Kusto Query Language, making threat hunting less of a wild goose chase.

READ THE ARTICLE ON THE REGISTER

Oxford University develops AI tool to track virus variants

Oxford University teamed up with the Harvard brainiacs to whip up this snazzy AI tool they've named EVEscape, and it’s pretty much a crystal ball for virus variants. They're saying it could've called the shots on the Covid-19 mutations, making it a potential game changer for cooking up vaccines that can roll with the punches when viruses decide to switch things up.

During the whole Covid hullabaloo, we saw the virus trying on new genetic outfits, leading to different waves of chaos. Some variants were sneakier than others, playing hide and seek with our immune system. Remember Omicron? Big on infection, not so much on the hospital drama.

EVEscape is like a blend of deep virus know-how and a smart algorithm that takes a stab at predicting a virus's next fashion statement. The team ran it through a time travel test, feeding it data from way back in February 2020, and voila! It called the Covid-19 mutations like Babe Ruth calling his shot. It even sniffed out which fancy antibody treatments would start striking out as the virus evolved.

READ THE ARTICLE ON BBC

Exploring the Boundaries of GPT-4 in Radiology

This paper is taking us on a wild ride through the land of GPT-4. This time, it’s dipping its virtual toes into the world of radiology reports, and the results? Not too shabby, if you ask me.

The team behind this paper decided to pit GPT-4 against the top dogs of radiology-specific models, just to see who comes out on top. And guess what? GPT-4 either kicked butt or held its own in most of the tests. With no prep, it already showed a decent improvement over the other models in understanding how sentences are related and making smart guesses about text.

When they fed GPT-4 some examples to learn from, it really started to shine, even matching the performance of models that had been trained specifically for this stuff. Not too shabby for a generalist, right?

READ THE ARTICLE ON HUGGING FACE

With the help of AI, cardiologists can predict who will develop A-Fib

Doctors in San Francisco found a shiny new way to predict if you're about to dance with A-Fib, a tricky heart rhythm issue. The human peepers? Not so great at catching all the details on an EKG. But machines? They're like Sherlock with a calculator.

Dr. Yuan and his techy pals fed a computer a boatload of EKG info, and the computer – bless its digital heart – started guessing who's on the A-Fib hit list. The more it guessed, the better it got. Think of it as training a dog, but this one predicts heart issues. The big news? The computer's guesses weren't half bad. It’s like a crystal ball for your ticker. So, watch out world, machines are teaming up with docs to keep our hearts playing the right beat.

READ THE ARTICLE ON NPR

Some tech leaders fear AI. ScaleAI is selling it to the military.

Alexandr Wang, the 26-year-old CEO of AI company ScaleAI, has set his sights on aiding the U.S. military in its battle against China. ScaleAI, known for organizing and labeling data to train AI algorithms, snagged a $249 million contract with the Department of Defense. Wang, a self-proclaimed "China hawk," believes that without AI, the U.S. will lose its technological edge.

Wang's company has been pioneering AI advancements, but it faces stiff competition from tech giants like Google, Microsoft, and Amazon, all vying for Pentagon contracts. ScaleAI has also faced criticism for its practices in the Global South.

Despite concerns about AI's role in warfare, Wang insists that rigorous testing is essential before deploying advanced AI on the battlefield. With geopolitical tensions rising, the tech industry is gravitating towards government partnerships, and ScaleAI aims to play a pivotal role in maintaining U.S. leadership in an increasingly complex world.

READ THE ARTICLE ON WASHINGTON POST

ChatGPT in Cars: DS Automobiles announces AI integration in its cars; check how it will work

DS Automobiles, part of the Stellantis gang, is getting in on the AI action by slapping ChatGPT, that chatty computer brain, into its cars. They're rolling this out across their DS 3, DS 4, DS 7, and DS 9 lineup.

ChatGPT is like the Sherlock Holmes of AI, soaking up all the knowledge from the internet to answer your burning questions. DS Automobiles is plugging this digital chatterbox into their DS IRIS SYSTEM. You can talk to your car by saying "OK IRIS" or pushing a button on the steering wheel. No need to take your eyes off the road or your mitts off the wheel.

Olivier François, Head of DS Automobiles, wants to make your car ride an unforgettable experience. And Yves Bonnefont, Stellantis Chief Software Officer, is amazed at how this chatty AI model can talk the talk. They're kicking off a pilot phase with "SoundHound AI powered by ChatGPT" and offering it for free for six months if you subscribe between October 19, 2023, and February 29, 2024.

READ THE ARTICLE ON BUSINESS TODAY

White House announces 31 tech hubs to focus on AI, clean energy and more

The White House just declared 31 tech hubs across the U.S. to help boost our tech game. Each of these hubs can compete for up to $75 million in grants. It's all part of President Biden's plan to amp up the tech sector and American competitiveness. These hubs will focus on stuff like quantum computing, AI, clean energy, medicine, and biotech. They're spread across 32 states and Puerto Rico, making sure not just the big cities get a piece of the pie. Biden's talking up his economic agenda, and these hubs are part of the pitch.

READ THE ARTICLE ON CNBC