OpenAI's Bold Move Against Catastrophic Risks

Today:

OpenAI forms team to study ‘catastrophic’ AI risks, including nuclear threats

OpenAI's whipping up a new crew called "Preparedness" to keep an eye on any scary AI risks that might pop up. Aleksander Madry, a big brain from MIT, is leading the charge. They're watching out for AIs that might trick us or even do some serious damage, like some real sci-fi stuff – think chemical or nuclear threats.

While OpenAI's boss, Sam Altman, has been pretty vocal about AI possibly going rogue, this move kinda ups the ante. OpenAI's also running a contest, looking for ideas on how AIs could be misused, with a fat $25k prize for the best pitches. The aim? Make sure that as AIs get smarter and more powerful, they're still on our side.

This announcement dropped during a big U.K. AI safety meeting, and it follows OpenAI's earlier move to study super-smart AIs that might be smarter than us. So yeah, they're taking this pretty seriously.

READ THE ARTICLE ON TECHCRUNCH

Google Maps unveils new, AI-powered navigation tools

Google Maps is stepping up its game with some fancy new AI-driven features. If you’re hitting the road, walking, or biking in a bunch of big cities, you can now get a sneak peek of your route with an AI preview. It’s like a virtual tour, showing you the lay of the land, bike paths, and even the traffic and weather you might hit.

On top of that, searching in Google Maps is getting sharper. Thanks to computer vision, it can dig up results based on pictures folks have shared, as well as the usual text stuff. So, if you’re on the hunt for a coffee shop with cool latte art, Maps has your back. And if you’re just wandering around, wondering what to do, the app’s new search categories will help sort things out for you.

Got an iOS or Android phone? You can now use Lens in Maps to get the lowdown on places just by pointing your camera at them. This works in a bunch of new cities, so you can spot things like ATMs or restrooms in a snap.

Drivers, you’re not left out. Google Maps is serving up more details on your drives, showing off buildings and making sure you know about those tricky highway lane changes and exits. Electric car folks, there’s something for you too. The app is making it easier to find charging stations, complete with all the specs you need to know.

READ THE ARTICLE ON FAST COMPANY

Microsoft reorgs its Xbox and marketing teams to prepare for an AI and gaming future

Microsoft is shaking things up in their Xbox and marketing teams, getting ready for a future full of AI and gaming. They’ve just bought Activision Blizzard and are now promoting Matt Booty and Sarah Bond to big roles in the Xbox world. Booty’s taking charge of a bigger chunk of the gaming scene, making sure all the different teams play nice together and avoid any hiccups like the Redfall situation. Bond is the new big boss of all things Xbox, handling everything from the gadgets you play on to the software that makes it all tick. She's been a big deal at Xbox since 2017 and even represented the company in court recently.

Microsoft is gearing up for a big push into gaming and AI. They’ve been working hard to be a frontrunner in the AI game, integrating it into everything from Windows to Office apps. Their new tool, Microsoft 365 Copilot, is set to launch next week, giving businesses a hand with everything from writing emails to analyzing data. The CEO, Satya Nadella, is super excited about the possibilities, comparing their new AI tool to the Start button on Windows, saying it's going to change the way we use our apps and get things done.

READ THE ARTICLE ON THE VERGE

Descript's big AI feature release: New AI voices, AI actions, and much more

Descript's rolling out a ton of new AI goodies, really upping their game from when they first hit the scene in 2018. They've been more than just a one-hit wonder with their Overdub feature; they've been cooking up quality tools, not just flashy tricks.

They’re diving into the AI feast, starting with a revved-up version of their in-house AI voice tool. They promise it’s like upgrading from a clunker to a rocket ship, with faster setup and top-notch quality that could let you create a whole voiceover without saying a word.

But that's just for starters. Over the next couple of months, they're dropping a whole series of new features, all designed to make audio and video editing as smooth as butter. They've got tools for suggesting chapter markers, summarizing scripts, writing promotional copy, and even an “Ask AI Anything” feature.

READ THE ARTICLE ON DESCRIPT

Luminar Neo brings generative AI tools to hobbyist photographers

Skylum's Luminar Neo is shaking things up for hobbyist photographers by rolling out easy-peasy generative AI tools. It's kind of like Adobe for the everyday Joe, but with a couple more tricks up its sleeve. They're giving folks a hand with erasing unwanted bits from photos and jazzing up their images without breaking a sweat. The software's got its roots in a bunch of iOS apps the founders whipped up back in the day, but now they're all about making photo editing a walk in the park.

Luminar Neo is mainly for the photography buffs, but some pros are catching on and using it alongside their fancier tools. They've got big plans to drop a new AI feature each month till the end of the year, all based on some pretty solid tech.

READ THE ARTICLE ON TECHCRUNCH

Shutterstock will now let you transform real photos using AI

Shutterstock's shaking things up, now letting folks use AI to edit and mess around with its huge pile of images. They've got some fancy new tools, like Magic Brush, for tweaking pictures just how you want by painting over them and saying what you want changed or nixed.

It's all still in the testing phase, but you can also whip up different versions of a picture, stretch out the background, and even get the AI to smartly resize images to fit whatever you need. There's also a nifty tool to drop the background out of a picture.

If you edit an artist's picture and then use it, Shutterstock will still throw some cash their way. But, if you're using AI-edited stuff, don’t even think about trying to license it on Shutterstock's site. They're clear: that's a no-go.

On top of all that, they're also giving their AI image maker, which they kicked off in January, a bit of a facelift with the latest from OpenAI's DALL-E. They've even locked in a deal to keep training DALL-E on their images for another six years.

READ THE ARTICLE ON THE VERGE

Outset is using GPT-4 to make user surveys better

Outset is shaking things up in the world of user surveys. Founded by former Deloitte strategist, Aaron Cannon, and his pal Michael Hess, Outset uses OpenAI's GPT-4 to automatically hold interviews and break down the results. The big idea? Make gathering feedback less of a headache.

Before, businesses would shell out big bucks and loads of hours for these insights. With Outset, even big names like WeightWatchers can get the lowdown from over 100 chats in just a day. They're growing fast too, with plans to hire more folks after raking in $3.8 million in funds. Despite some bumps in the economy, Outset's future looks bright with their smart use of AI.

READ THE ARTICLE ON TECHCRUNCH

The Beatles: ‘final’ song Now and Then to be released thanks to AI technology

The Beatles are dropping a "final" track, "Now and Then," with all the Fab Four playing on it, thanks to some nifty AI tech. Paul McCartney is over the moon, saying it's like the old days with John Lennon’s voice sounding crystal clear. The song, originally belted out by Lennon in his NYC Dakota building pad back in the 70s, was gifted to McCartney by Yoko Ono in '94. But tech hiccups back then meant it got shelved while other Lennon demos got the Beatles treatment.

Despite initial jitters about AI messing with Lennon's performance, McCartney assures everyone it's all legit. He’s just stoked to be jamming Beatles-style in 2023. The song's getting a sneak peek in a short doc, then a full reveal and a single release, paired with “Love Me Do.” It's a big deal, especially since George Harrison wasn't a fan back in the day. Yet, Olivia Harrison thinks he'd be all in now. Lennon’s son, Sean, finds the whole thing pretty surreal and touching.

READ THE ARTICLE ON THE GUARDIAN

Forbes now has its own AI search engine

Forbes just rolled out a new AI search engine named Adelaide, giving a nod to the founder’s wife. It’s still in the testing phase, but it's all about dishing out personalized article recommendations and quick answers to readers’ questions. The catch? It's gotta be something Forbes has covered in the last year.

Vadim Supitskiy from Forbes is all in, hoping this techy tool will get more folks interacting with their content. Readers can hit up Adelaide on the website, throw a question its way like “Who’s the top dog in Nebraska when it comes to cash?” and bam—there’s a summary about Warren Buffett plus some related reads. And the cool part? You can keep the chat going and Adelaide keeps up, serving more info based on your back-and-forth.

Other news outlets are hopping on the AI bandwagon too, with their own chatbots. But Adelaide's not Forbes' first rodeo with AI; they've been playing around with tools to help their writers since 2019. And they've got big dreams for Adelaide, aiming to give it a memory boost with access to their whole archive, all the way back to 1917.

READ THE ARTICLE ON FORBES

Meta says users and businesses have 600 million chats on its platforms every day

Meta's going all in on business messaging to make some extra dough. On a recent call about how they did in Q3 2023, Mark Zuckerberg dropped that people and businesses are chit-chatting over 600 million times daily on their platforms. He didn't really spill the beans on what types of messages these are, but mentioned that a ton of folks in India – especially on WhatsApp – are messaging businesses.

Lately, WhatsApp's been rolling out new tools for businesses to better connect with their customers. Plus, Meta's looking into using fancy AI tech to let business accounts chat with customers about buying stuff or getting help. Zuckerberg thinks this AI thing's gonna be huge, especially for places where it's pricey to have real folks answering customer questions.

READ THE ARTICLE ON TECHCRUNCH

New Training Method Helps AI Generalize like People Do

Researchers found a cool new way to train AI that might make it think a bit more like us humans. Instead of stuffing them with tons of info, it's all about how they learn. You know how we mix ingredients to make bomb recipes or string words together to make sense? That's called "compositionality." Well, AI has had a tough time nailing that.

Till now, even the big dogs like GPT-4 sometimes goof up, especially in long convos. But these brainy folks came up with a new way to train AI using a made-up language of random words and colorful dots. The AI learned this language and even bested humans in some tests! And get this, it wasn't some mega AI; it was a small one, but trained the right way.

READ THE ARTICLE ON SCIENTIFIC AMERICAN

This Startup Raised $30 Million To Take On Microsoft's AI Notetaker For Doctors

Startup Abridge just got $30 million to tackle Microsoft's AI notetaker for docs. Abridge, now valued at $200 million, is going head-to-head with Microsoft's Nuance, a big dog in the game.

Abridge's tool? An app that lets doctors talk during patient visits. This AI-powered gadget then turns that chatter into a neat note, minus the fluff. No need for humans to be involved. It's a big time-saver for docs. They can then shoot the note straight into a patient's electronic health record.

READ THE ARTICLE ON FORBES

China accepts invitation to AI summit in Britain - Deputy UK PM

China's accepted an invite to a big AI summit in Britain, where the UK's Deputy PM and other bigwigs from over 40 countries are gonna talk about everything AI—from making sure it's on the up and up, to using it for good, and figuring out the rules of the road. The summit’s happening in mid-November in London.

The main goal? Get everyone on the same page about how to handle AI, making sure it’s used responsibly and ethically. They're planning to talk about a whole bunch of stuff like keeping AI in line with human values, making it trustworthy and inclusive, and balancing innovation with the rules. Plus, they want to figure out how to work together on setting global AI standards.

Now, China’s a major player in AI, pouring a ton of money into it and cranking out a bunch of AI patents and startups. They've got big plans to lead the world in AI by 2030. So, it’s no shocker they said yes to the summit. They want to strut their stuff, connect with other countries, and push their own ideas on how to handle AI, emphasizing sovereignty and cooperation.

READ THE ARTICLE ON MEDIUM

Ilya Sutskever, OpenAI’s Chief Scientist, On His Hopes And Fears For The Future Of AI

Ilya Sutskever, the big brain behind OpenAI, met up in San Francisco to dish on AI's future. Instead of cranking out more AI like ChatGPT or DALL-E, he's now dead-set on making sure future super smart AI doesn’t turn on us. He's got some out-there beliefs, like ChatGPT maybe having a hint of consciousness, or folks one day choosing to become part machine.

Despite sounding sci-fi, these ideas aren't as wild as you'd think, especially since ChatGPT came on the scene and shook things up. OpenAI is the talk of the town, and while CEO Sam Altman is out mingling, Sutskever mostly lays low. When it comes to AI, though, he's all in, saying it's gonna be huge.

Sutskever isn't new to the AI game. Born in Russia but raised in Jerusalem, he later studied in Canada with AI legend Geoffrey Hinton. Sutskever and crew made waves with neural networks, especially after designing AlexNet in 2012.

This big breakthrough was like lighting the AI fireworks. After a short gig at Google, Sutskever joined the newly-founded OpenAI in 2014, where he’s been pushing the boundaries of what's possible. Even though OpenAI had a rocky start, Sutskever always believed in the power of deep learning, and boy, has it paid off!

READ THE ARTICLE ON MIT

5 Industries Where AI Brings Opportunities For Innovative Startups

This article is all about how AI, or artificial intelligence, is shaking things up in a bunch of different fields. And now's the time to jump in, especially for startups, because the whole AI scene is hot and there's a lot of opportunity to grow and make a mark.

• HEALTHCARE: It's helping doctors and nurses get better at diagnosing and treating patients, especially with all the data they have to deal with. A standout example here is PathAI, which is helping pathologists nail their diagnoses and provide top-notch care to patients.

• FINANCE: AI is making things run smoother and faster, from trading stocks to helping customers. And the data? There's a ton of it, which is like a playground for AI. Take Robinhood, for example. They're using AI to give folks personalized investment advice, making the whole investing thing a lot more user-friendly.

• EDUCATION: AI is making learning more tailored to each student's needs, whether it's through online platforms or in the classroom. And Duolingo? They're knocking it out of the park by using AI to give people lessons in new languages that are just right for them.

• ENERGY: The energy sector's using it to keep equipment running smoothly and use resources more wisely. It's all about cutting down on waste and saving money, and startups like Uptake are leading the charge with their AI tools for maintenance.

• AGRICULTURE: Lastly, AI is helping farmers get more bang for their buck by making farming super precise. We're talking about monitoring crops, figuring out the best time to plant and harvest, and cutting down on waste. Blue River Technology is a big player here, with their "See & Spray" solution that treats plants individually, saving on herbicides and boosting crop yield.

So in a nutshell, AI's opening up a world of possibilities across the board, from helping patients get better care to making farming more efficient. And for startups, the time is ripe to dive in and ride the AI wave.

READ THE ARTICLE ON FORBES