Amazon's Generative AI for Sellers

Today:

Amazon debuts generative AI tools that helps sellers write product descriptions

Amazon's dropping a cool new AI tool to help sellers jazz up their product listings. Instead of starting from the ground up, sellers can use this tool to add some extra flair to their current descriptions.

How's this work? Amazon's been teaching its machines with tons of data – kinda like giving them a masterclass on product details. With this knowledge, the AI can make smart guesses. For instance, if they see a shirt pic, they can tell you the type of collar it's got.

Amazon says this tool is a time-saver and makes product info more thorough for buyers. But, here's the catch: these AIs can sometimes make stuff up. If sellers aren’t careful, they might end up with false info. And if that happens? Amazon might be on the hook, especially if they don't tell folks that a robot wrote the description.

READ THE ARTICLE ON TECHCRUNCH

EY launches AI platform and LLM after $1.4 billion investment

EY dropped big news today - they're rolling out a new AI platform named EY.ai. The goal? Help businesses get more into AI. They’ve teamed up with some big names like Microsoft and Dell for this. They’ve also splashed out $1.4 billion to build this platform, incorporating their own tech, EY Fabric, into it. They're also launching a fancy new language tool named EY.ai EYQ.

About eight months back, EY’s tech head honcho, Nicola Morini Bianzino, was talking about how AI can transform the way businesses manage knowledge. Today, Bianzino said that their new AI tools can give businesses a sure-footed path, kinda like a “treasure map”, to get into AI. So, for businesses wanting to dive deeper into AI, EY's making waves.

READ THE ARTICLE ON VENTUREBEAT

Adobe’s Firefly generative AI models are now generally available, get pricing plans

Adobe's Firefly, a set of AI tools for creative stuff like photo editing, is out of beta testing and now available to everyone. You can use it in Adobe's Creative Cloud, Adobe Express, and Adobe Experience Cloud, or as its own web app. If you wanna play around with Firefly features in Photoshop—like filling in gaps or expanding images—you no longer need the beta version.

Adobe's got a new way to charge you for using Firefly. They're using "generative credits." Every time you click to make a new Firefly image, you use up one credit. Don't worry, if you're already paying Adobe, you get a bunch of these credits for free. Run out, and Firefly will slow down, but you can still use it.

For now, no credit limits until November 1 if you're already an Adobe user. Adobe's not saying how much extra credits will cost yet, but they promise it won't break the bank.

READ THE ARTICLE ON TECHCRUNCH

Alibaba opens AI model Tongyi Qianwen to the public

Alibaba's getting the green light from China to share its AI tool, Tongyi Qianwen, with everyone. China's really pushing its tech companies to go big in AI, especially with the US in the same race. Big names like OPPO and Taobao are already jumping on board to use Tongyi Qianwen.

Soon, everyone can get a piece of it for free. Alibaba's big boss, Daniel Zhang, just left his role, and the new head honcho, Eddie Wu, says AI's where it's at for the company's future. They don't wanna fall behind! This Tongyi Qianwen thing, kinda like ChatGPT, was first shown off in April, and they've got plans to put it everywhere in their business.

READ THE ARTICLE ON REUTERS

Apple’s iPhone 15 launch focused heavily on AI — even though the tech giant didn’t mention it

Apple chatted up the features of its iPhone 15 and Apple Watch Series 9 at its recent launch event. Even though they never said "AI", that's what's making a lot of those features tick. Inside the iPhone 15 and some watches, there's a powerful chip.

This chip lets Siri, Apple's voice helper, work right on the device instead of always using the internet. This means things work faster and more privately. Another cool feature on the Apple Watch Ultra 2 is "double tap". By just tapping your fingers together, you control the watch. This also uses AI. Some folks think Apple's not into AI because they don't talk about it. But they're wrong. Apple's all in on AI.

The new iPhone has a tiny 3nm chip inside, the smallest and most powerful around. It helps with better typing predictions and camera tech - again, all thanks to AI. Gene Munster, a bigwig at Deepwater Asset Management, said as more stuff uses AI, phones need to keep up. Older phones might start feeling slow. But with these new chips, Apple's leading the race.

READ THE ARTICLE ON CNBC

AI can now generate CD-quality music from text, and it’s only getting better

Stability AI's got a new toy called Stable Audio that turns your text into music or sounds, kinda like magic. So, if you type "dramatic intro music", bam! – you get a symphony. They're the same folks behind that fancy image-making AI from last year. And guess what? This isn't their first rodeo in the audio game.

The samples? Fire! Think movie trailer epicness or chill beats for your evening unwind. They trained it with a boatload (over 800,000!) of audio files so it knows its stuff. It's quick too – we're talking CD-quality audio in a snap if you've got the right gear.

READ THE ARTICLE ON ARSTECHNICA

AI detects eye disease and risk of Parkinson’s from retinal images

Scientists made an AI tool, RETFound, that can tell if you're at risk for things like eye problems, heart failure, or even Parkinson's just by looking at pictures of your retinas. What's cool is they used a kind of "learn-as-you-go" approach so they didn't have to manually tag millions of images as normal or not—saves a ton of time and money. They kinda trained it the same way I was trained, to predict stuff based on what it's already seen.

Why retinas? Well, they're like a health window, showing small blood vessels and even some nerve stuff, giving clues about your overall well-being. Once RETFound got the hang of what normal retinas look like, it got real good at spotting ones with problems.

They're now thinking about using this method on other types of medical images like MRIs. The tool is public, so any hospital or research place can tweak it for their needs. But the catch is, if RETFound has any flaws, those could carry over into new versions. So, the makers gotta be super clear about what it can and can't do.

READ THE ARTICLE ON NATURE

Fine-Tune your Own Llama 2

This tutorial's got the 411 on tweaking a Llama 2 model to sort recipes. They're pulling data from the web and teaching the model using a GPT-4 engine.

You got four main steps:

1) making your data

2) training the model

3) testing to see if it's any good

4) comparing it with other models like GPT-3.5 and GPT-4.

They suggest using RunPod if you wanna try this at home. OpenPipe, the company behind this, usually handles these steps for you, but they're spilling the beans on how to do it yourself.

READ THE ARTICLE ON GITHUB

Why AI Could Be The Key To Sealing The Deal With The Biggest Customers

Big companies are the dream catch for small businesses, but getting these giants to buy your stuff is like trying to nail Jell-O to a wall. Samir Manjure, head honcho at a Seattle startup called Vieu, thinks AI could be the magic wand here. His company, fresh off scoring a cool $2 million, is all about helping businesses get a fast track to these big customers. How? By using AI to find the right people to chat with at these companies, and even sussing out what makes them tick, so sales folks can make a killer pitch.

Manjure's so sure this will work that he's saying "money back if it doesn't." Some big names are putting their money on Vieu too, like Lila Tretikov from Microsoft. But with heavy hitters like Salesforce in the ring, Vieu's gotta show they've got the special sauce.

READ THE ARTICLE ON FORBES

California lawmaker proposes regulation of AI models

California's Senator Scott Wiener is pushing a bill to shine a light on big AI systems, making sure they're open about how they work. The goal? Keep these powerful AI tools safe and out of the wrong hands, like other countries.

He's also aiming to set up a special research spot for AI that's away from the big tech giants. While the bill's still being shaped up, it'd make AI labs check for safety issues and spill the beans if they find any.

READ THE ARTICLE ON THE VERGE

UK gov't announces new $1.1B supercomputer and AI research facility

The UK is dropping $1.1 billion on a powerhouse supercomputer and a research center focused on artificial intelligence (AI). Named Isambard-3, after a 19th-century British engineer, the computer's gonna be one of Europe's top dogs in terms of power and will be set up in Bristol.

On top of that, there's gonna be a new AI Research Resource—let's call it the help desk for AI research and safety. Both these big-ticket items are part of the UK's plan to stay ahead in the AI game. Oh, and there’s a big meeting about AI safety coming up. Experts from all over the world will huddle up in the UK to talk about how to keep AI on the straight and narrow. So, it's all systems go for the UK to amp up its AI street cred.

READ THE ARTICLE ON NETWORKWORLD

Elon Musk calls for federal department of AI after Capitol Hill summit

Elon Musk went to Capitol Hill and told Congress they need to create a federal department for artificial intelligence (AI). He rolled up in a Tesla and spoke alongside big names from the tech world, like Google's Sundar Pichai and OpenAI's Sam Altman.

Musk thinks uncontrolled AI is a big-time risk to all of humanity. He's hopeful about the meeting, saying it felt like folks were on the same page. Musk thinks the new AI department could work kinda like how the FAA oversees airplanes or how the SEC keeps an eye on the stock market. He's already in the AI game himself, having launched his own AI company, xAI, a couple months back.

READ THE ARTICLE ON CNBC