Amazon's New AI Image Generator for Advertisers

Today:

Amazon rolls out AI-powered image generation to help advertisers deliver a better ad experience for customers

Amazon's throwing a lifeline to advertisers up to their eyeballs in creative struggles. Their new toy? AI-powered image magic that transforms basic product shots into eye-catching lifestyle ads.

They're dishing out a beta version of their generative AI tool to help brands whip up ads that don’t just blend into the background. Picture this: a toaster. Now picture it on a kitchen counter, hanging out next to a croissant. Which one's making you click? Exactly.

The big guns at Amazon say it's all about cutting the hassle for advertisers, big and small, and spicing up the customer experience. No tech wizardry required. Just pick your product, hit GENERATE, and watch the AI do its thing, dishing out lifestyle shots faster than you can say “increase my click-through rate.” Want to tweak it? Throw in a short text prompt and you're golden.

READ THE ARTICLE ON AMAZON

Anthropic, Google, Microsoft and OpenAI announce Executive Director of the Frontier Model Forum and over $10 million for a new AI Safety Fund

Big tech names like Google, Microsoft, OpenAI, and Anthropic are teaming up and throwing more than $10 mil into a fancy-schmancy "AI Safety Fund." Why? To make sure we're not blindly driving AI into a wall. They’ve tapped Chris Meserole, a guy who knows a thing or two about tech policy, to head the Frontier Model Forum. This group’s gig? Ensure AI develops safely and responsibly on a global scale.

Oh, and they’ve got some big-name donors backing this fund. Think folks like the Patrick J. McGovern Foundation, Eric Schmidt, and Jaan Tallinn, just to name a few. And they’re hoping for more to jump on this bandwagon.

It's gonna support smarty-pants researchers around the world, focusing on "red teaming." They're gearing up to launch their first call for proposals, and Meridian Institute is going to manage the fund's purse strings.

They want to set common definitions and best practices to make AI safer. They're even cooking up a system to share info when they find weak spots in AI tech. What's next on their to-do list? They're setting up an Advisory Board, planning to dish out some grants, and will be dropping more tech insights.

READ THE ARTICLE ON GOOGLE

Microsoft Business Applications Launch Event introduces wave of new AI-powered capabilities for Dynamics 365 and Power Platform

Microsoft Dynamics 365 and Power Platform are getting a glow-up with a bunch of new features rolling out from October 2023 to March 2024. We’re talking hundreds of updates, people!

Now, they’ve thrown in some fancy AI capabilities into Copilot, and apparently, it's the talk of the town with over 130,000 organizations giving it a whirl. This thing is supposed to make your work life a breeze, helping you whip up insights, save precious time, and sprinkle some creativity here and there.

They showcased how some big shots like Nestlé and Domino’s Pizza are jumping on this bandwagon and transforming their businesses. And if you missed the live show, no worries. They've got it all recorded for your viewing pleasure. Plus, there’s more geeky goodness coming up at Microsoft Ignite 2023.

READ THE ARTICLE ON MICROSOFT

The much-hyped AI device created by ex-Apple designers will run GPT-4 and have a 'Trust Light' that lets people know when it's recording

Humane, a startup by two Apple escapees, is about to release their latest gizmo: the "Ai Pin". Think of it as your trusty personal assistant you can wear, and yep, it's got the brains of GPT-4 running inside.

This baby's got a "Trust Light" that tells folks around you when it's recording. No sneaky business here! And you don't need to shout "Hey, Pin!" because there's no wake word. Just clip it on your shirt or pants and you're good to go. It even made a cameo at a Paris fashion show. Talk about trendy!

READ THE ARTICLE ON BUSINESS INSIDER

Google will require Android apps to better moderate AI-generated content

Google's putting the squeeze on Android apps that spit out AI-made content. If your app's got AI churning out words, pics, or sounds, you better slap on a "report" button for any funky business by next year. And this ain't just for any app; we're talking about chatbots, picture-making apps, and ones that cook up voice or video using AI. But if your app's just giving a short-n-sweet of Moby Dick using AI? No worries, you're good.

Oh, and if your app's nosy about photos and videos? Google's cutting down on the snooping. Only apps that really need all your pics will get to see 'em. Others? They get a peephole, not a panoramic view. Because, y'know, privacy and all that jazz.

READ THE ARTICLE ON THE VERGE

Jina AI Launches World's First Open-Source 8K Text Embedding, Rivaling OpenAI

Jina AI, a smarty-pants AI company from Berlin, just threw down the gauntlet and launched the world's first open-source 8K text embedding model. Think of it as teaching a computer to understand really, really long texts. They've named this beast "JINA-EMBEDDINGS-V2." Now, these Berliners are pretty chuffed because their new toy matches up nicely with OpenAI's fancy-schmancy proprietary model in a head-to-head competition.

Here's the scorecard: While OpenAI's model still holds its own, Jina's model nudges ahead in a few categories. It's kinda like comparing two top draft picks — one might run faster, but the other might catch better.

READ THE ARTICLE ON JINA.AI

Google considers subscription business models for its new AI offerings

Google's big cheese, Sundar Pichai, is playing with the idea of making folks pay for some fancy new AI products, like a jazzed-up search engine and a chatty robot that could give ChatGPT a run for its money.

Now, Wall Street's getting antsy, wondering what's the deal and how this is gonna pay off. Google's bread and butter has always been ads—especially in search. So messing with that could be like poking a sleeping bear.

Pichai, though, seems pretty stoked about how folks are taking to their new AI toys. And he's dropping hints that maybe, just maybe, they'll start charging for some of this stuff, kind of like how YouTube's got folks coughing up cash for extra goodies.

READ THE ARTICLE ON BUSINESS INSIDER

The White House will reportedly reveal a ‘sweeping’ AI executive order on October 30

The White House is gearing up to drop some major AI news. Word on the street is that next Monday, Biden's team will whip out a big ol' executive order on artificial intelligence. What's the game plan? Making sure Uncle Sam checks AI models real good before government agencies play with 'em. And get this – they're even thinking of rolling out the welcome mat for smarty-pants immigrants, a move that was kinda on lockdown during Trump’s time.

The big departments, you know, Defense, Energy, and those sneaky intelligence folks, gotta brainstorm how to sprinkle some AI magic in their daily grind. And oh, they're all about beefing up cyber protection. Can't be too safe in the digital wild west!

READ THE ARTICLE ON ENGADGET

AI ‘breakthrough’: neural net has human-like ability to generalize language

Scientists made this fancy computer brain that's got a knack for language like we do. It can take new words, throw them in its word salad, and still make sense—something us folks call "generalizing."

You know ChatGPT, that chatty bot that sounds like your neighbor Joe? Turns out, it's not as good at this wordy jazz as the new tech or even regular Joes and Janes. Published in "Nature", this could mean our future robots might chat with us more like a buddy at a bar, minus the oddball mistakes.

Paul from Johns Hopkins says this is a big step up. Here's the thing: humans learn a word and can use it everywhere. Learn "photobomb," and you're talking about "photobombing" your grandma's Zoom bingo night. But computers? They usually need tons of examples before they get it. For 40 years, smart folks have debated if computer brains can really get how we think.

To see who's boss, the brains behind the study tested folks on fake words. They linked words like "dax" to colors (think: red means "dax"). They then tested if folks could use more complicated rules. Spoiler: humans did swell, getting it right about 80% of the time.

Then they gave the new computer brain the same pop quiz. They taught it to think more like humans, including making our silly mistakes. And guess what? This tech wiz nailed it, sometimes even outdoing the humans.

READ THE ARTICLE ON NATURE

AI could help doctors make better diagnoses

Dr. Michael Mansour, an infectious disease expert at Massachusetts General, is giving AI tech a whirl to help make medical calls. The guy’s big on tackling gnarly mold and yeast issues in transplant folks. He's using this tool, UpToDate, think of it like Google but for docs, to dig through heaps of articles and research.

Mansour shared a "what if" story: patient from Hawaii, unknown bug. He types in symptoms, gets a laundry list of potential culprits from dengue to jellyfish stings. But it's like finding a needle in a haystack. He reckons this fancy AI could cut through the clutter.

Wolters Kluwer Health, the brains behind UpToDate, is jazzing it up with AI so docs and the database can chit-chat. Dr. Peter Bonis, their head doc, wants to make it as good as asking a top-notch clinician.

READ THE ARTICLE ON NPR

Bill Gates feels Generative AI has plateaued, says GPT-5 will not be any better

Bill Gates reckons the AI hype train might be slowing down a bit. While everyone's been raving about ChatGPT, and OpenAI’s language model has been the talk of the town, Gates thinks the tech might have hit its peak. He recently told a German paper that, even though the jump from GPT-2 to GPT-4 was "wild", he doesn't have high hopes for GPT-5 outdoing its older sibling. But, plot twist: he did admit he could be off the mark on this one.

He's also jazzed about AI helping out in the developing world, especially dishing out health tips via phones. On the money front, while it costs a small fortune to train big AI models (with some chips setting you back a cool $30k), using them is getting cheaper by the day.

Gates is optimistic about AI speeding up drug and vaccine development and sees it playing a bigger role in healthcare. He's also curious about the "black box" nature of AI and thinks cracking that code will be a game-changer. And for those waiting on Artificial General Intelligence (AGI)? Gates says, don't hold your breath – but when it does show up, it'll be a big deal.

READ THE ARTICLE ON INDIAN EXPRESS

AI-created child sexual abuse images ‘threaten to overwhelm internet’

Alright, buckle up because this one's a doozy. The Internet Watch Foundation (IWF), kind of like the internet’s watchdog, just dropped a bombshell: nearly 3,000 AI-generated child sexual abuse images floating around that break UK law. We're talking serious nightmare fuel here.

So here’s the lowdown: the bad guys are using AI to cook up new, horrific images, sometimes messing with pics of real abuse victims or even giving celebrities a creepy digital facelift to make them look like kids. They're not stopping there—they're also "nudifying" pictures of kids in clothes. Just when you thought the internet couldn't get any darker.

The IWF has been on this case for a while, waving red flags since summer. But now, they're saying things are ramping up fast. Susie Hargreaves, the big boss at IWF, put it bluntly: “Our worst nightmares have come true”. She's worried the fakes are getting so good, they're almost indistinguishable from the real, tragic deal.

READ THE ARTICLE ON THE GUARDIAN

[Long read] Deep dive into AutoGPT: A comprehensive and in-depth step-by-step guide to how it works

The article kicks off with the author, knee-deep in the AI world, tripping over AutoGPT and deciding to give it a whirl. Their method? Roll up the sleeves and dive in, tinkering around and taking copious notes about what goes in and what pops out.

The nitty-gritty of the article breaks down the play-by-play of what happens behind the AutoGPT curtain. The author lays out the ingredients of the experiment—inputs, agent settings, prompts, and outputs—and then cooks up observations that are as insightful as they are concise.

READ THE ARTICLE ON AIRT.AI