AI Chip Showdown: Microsoft's Debut and OpenAI's Ambitions Beyond Nvidia

Today:

Microsoft to Debut AI Chip Next Month That Could Cut Nvidia GPU Costs

Microsoft's dropping some tech news next month! They're about to show off their own AI chip at their big yearly meeting. Why's this a big deal? Well, right now they're using Nvidia's chips for all their fancy AI stuff, but those are hard to get 'cause everyone wants them.

Their new chip is like Nvidia's, mainly for those big computers that make things like chatbots work. If this chip does its job, Microsoft might not need Nvidia's stuff as much. That's good news for folks who use Microsoft's cloud services and apps.

READ THE ARTICLE ON THE INFORMATION

ChatGPT-owner OpenAI is exploring making its own AI chips

OpenAI, the brains behind ChatGPT, is toying with the idea of making their own AI chips. Word on the street is they even looked into buying another company to help out. They haven't made a solid move yet, but they've been chatting about how to deal with the pricey AI chip shortage they're facing. Right now, they rely a lot on Nvidia, but they're thinking of spreading their wings to other suppliers too.

CEO Sam Altman is really pushing to get more chips. Dude's been vocal about the lack of graphics processing units, especially since Nvidia pretty much rules that market. Sam's big worries? Not enough high-tech processors for OpenAI's software and the crazy high costs of running their gear.

To give you a sense, running ChatGPT ain't cheap. If it grows big, like Google search big, it could cost a whopping $16 billion a year in chips. Yikes.

If OpenAI jumps into the custom chip game, they'd join other tech giants like Google and Amazon. But making their own chip is a big deal and could cost them a ton. Even buying a chip company, while it might speed things up, isn't a guaranteed win. And any move they make will take time, so they'd still be leaning on big names like Nvidia for a while.

READ THE ARTICLE ON REUTERS

Meta Is Paying Creators Millions for AI Chatbots

Meta's shelling out big bucks to celebs like Tom Brady and Paris Hilton to turn them into chatbots. They're dropping over a million dollars for each celeb to let them use their faces and voices for these digital pals. Some big names are pocketing around $5 million for just a few hours of studio time. And guess what? These stars can still sell their digital selves to other AI gigs if they want.

At a recent event, Meta's head honcho, Mark Zuckerberg, showed off these new AI buddies that can chat with you or even help whip up a dinner recipe. And it ain't just about answering questions; it's about having a good time and connecting with folks. For instance, Brady's bot is a sporty jokester, while Paris Hilton's digs into mysteries.

With these moves, Meta's hoping to pull in the young crowd that's been ditching Facebook for TikTok. They're ramping up their AI game, recently launching a new tech called Llama 2. And soon, these chatbots might even talk back, not just text.

READ THE ARTICLE ON THE INFORMATION

Snapchat: Snap AI chatbot 'may risk children's privacy'

Snapchat's in hot water over its AI chatbot, with UK's top data hounds suggesting it might not be safe for kids. The UK's watchdogs say Snapchat didn't really think through the privacy stuff before letting this AI loose, especially for teens. If Snapchat doesn't clean up its act, UK might pull the plug on this feature.

This chatbot, called "My AI", is like a digital buddy users can chat with. But there's been some flak about how clear Snapchat is about the info this chatbot sees, like where users are.

Snapchat says they only peep your location if you say it's cool. But the big question is, do young Snap users really get what's at stake with their data? If Snapchat messes up big time, they could be slapped with a fat fine. For now, they're playing nice and saying they'll work with the UK to make things right.

READ THE ARTICLE ON BBC

The BBC is blocking OpenAI data scraping but is open to AI-powered journalism

BBC's looking into AI for journalism, but ain't cool with folks scraping their content. They've got 3 main rules: doing what's best for the public, having mad respect for artists' rights, and being straight-up about AI work.

BBC's teaming up with tech giants to keep news trusty. They've got plans to dig into how AI might change their game but kept it hush-hush in an email to The Verge. While the AP's shared stories with OpenAI, BBC's blocking OpenAI from snooping on their sites, saying it's to keep their content legit for the public. They're joining big names like CNN and NYT in this move.

READ THE ARTICLE ON THE VERGE

Salesforce IT injects generative AI to ease its massive datacenter migration

Salesforce is making a big move, switching its data centers from CentOS to Red Hat Enterprise Linux, and they're using their own AI to make it happen. That's a whopping 200,000 servers they're shifting! They're betting on this change for a sturdier, more flexible system.

Since CentOS is now old news and linked with Red Hat anyway, the switch to Red Hat makes sense. Plus, Salesforce isn't just doing this the old-fashioned way. They're using their latest tech, the Einstein 1 platform, to juice up the migration and get a better grip on their customer data. Tyson Lutz, the big shot overseeing all this, thinks this is a game-changer for how companies manage their tech gear.

READ THE ARTICLE ON CIO

Slingshot Aerospace harnessing AI to track suspicious satellites

Slingshot Aerospace, a company that uses AI to analyze space data, has sounded the alarm on Russia's new spy satellite, the Luch-2. Just like its older sibling, the Luch-1, which raised eyebrows in 2015 by hanging out between two commercial satellites, this newbie is pulling similar stunts in space.

Slingshot's tech noticed the Luch-2 making moves in space that were a bit fishy. From late September, the satellite drifted west about 1 degree daily. It then took a chill pill on October 2nd to hover around another bunch of satellites.

Audrey Schaffer from Slingshot mentioned they weren’t specifically watching this Russian satellite. Their software just caught it acting kinda sus. While she didn't spill the beans on which satellites the Luch-2 might be nosing around, she did say that this info is a big deal for anyone worried about space security. Just because this satellite isn't about to bump into anything doesn't mean it isn't a potential spy problem.

READ THE ARTICLE ON SPACE NEWS

The next frontier in biomedicine: AI models of the human immune system

Hey, remember how the Covid-19 pandemic threw us for a loop? It turns out we don't know much about our own immune system, even after whipping up vaccines in record time. But here's the kicker: it's not just about Covid. Many diseases, like HIV and even cancer treatments, stump us in the same way.

But there's a new sheriff in town: AI (Artificial Intelligence). Think about how we unlocked our genetic code (the human genome) years ago for billions of dollars. Now, it's like buying a pizza. Cheap and fast! With AI, we're diving deep into our immune system, which is WAY more intricate than our genes.

This isn't just nerdy science talk. We're talking about a world where we stop outbreaks before they start, whip up vaccines in a snap, and get a grip on heavy hitters like cancer and Alzheimer's.

READ THE ARTICLE ON STAT NEWS