- NATURAL 20
- Posts
- How Set Up Your OpenAI Store | Publish Your First GPT and Getting Verified as a Builder.
How Set Up Your OpenAI Store | Publish Your First GPT and Getting Verified as a Builder.
Learn how easy it is to develop and monetize GPT applications, even without technical expertise
Wes Roth
November 15, 2023
Today:
How Set Up Your OpenAI Store | Publish Your First GPT and Getting Verified as a Builder.
OpenAI's rolling out something cool called a GPT Store. You can make apps with their top-notch AI, no coding skills needed, and make some dough selling them. The store's gonna feature these creations and even highlight the best ones in different categories like productivity, education, or just for fun.
This GPT Store's a big opportunity for everyone, not just the tech wizards. You don't need to be a coding pro to jump in. What's key here is building your rep as a "Builder" – that's what they call the developers. You wanna create a bunch of related products, sort of like a family of apps that help each other out.
AI outperforms conventional weather forecasting methods for first time
AI just knocked it out of the park in the weather game, beating the old-school methods for the first time. Google DeepMind's new AI model, GraphCast, is now the champ at predicting weather up to 10 days out. This thing crushed it in tests, beating the top dog, the European Centre for Medium-range Weather Forecasts, in almost all the ways weather gets measured.
Even the ECMWF folks, who've been in the weather biz for ages, are giving props to GraphCast. They're saying it's not just better than other AI attempts but also more on point than their own system. What's wild is that GraphCast learned its skills by studying over 40 years of weather data.
Here's how it works: GraphCast looks at the current weather and what it was like six hours ago, then whips up a 10-day forecast in a snap, way faster and cheaper than the old method, which takes hours and a ton of energy. For example, GraphCast nailed the path of Hurricane Lee in the Atlantic, giving folks three extra days to get ready compared to traditional methods.
The big brains at ECMWF are thinking about mixing their old-school methods with this new AI magic, hoping to get the best of both worlds. The UK Met Office is also jumping on the AI bandwagon, working to blend AI with their existing tech, especially to keep up with the wild weather changes happening because of climate change.
YouTube will label AI-generated videos that look real
YouTube's gonna start giving heads up to folks when they're watching a video made with artificial intelligence. This means creators gotta be upfront about using AI or other digital magic to make their videos look real or else they could get kicked off the platform or lose their ad money. This new rule is coming soon.
What's more, YouTube's gonna let people ask for videos to be taken down if they use AI to fake someone recognizable. There's been a lot of buzz about AI tech that can make super realistic images and videos, called "deepfakes." Some folks are worried these could be used to trick people, like showing stuff that never happened or putting words in someone's mouth.
YouTube already said political ads with AI gotta be marked. This new policy they just dropped is an extension of that. YouTube's already against videos that are digitally tweaked to mislead people, especially if they could cause big trouble. But now, they're focusing on AI videos that could fool viewers.
YouTube Co-Founder Hurley’s Stealth AI Startup Raises Money from A-Star Capital, Ron Conway
Chad Hurley, the guy who helped start YouTube back in the day, is at it again with a fresh startup named EyeTell. This new gig is all about making quick, catchy videos using AI smarts. EyeTell is pretty new, just five months old, but it's already grabbing attention and some serious cash from big-time investors, though they're keeping the exact numbers under wraps.
Hurley's no stranger to the tech game. After YouTube, which Google bought for a jaw-dropping $1.65 billion in 2006, he's been dabbling in various projects, like another video app and owning slices of sports teams. He's also been big on domains, telling folks to grab '.com's' like they're hotcakes.
Before YouTube, Hurley was with eBay's PayPal, and even whipped up PayPal's first logo. Post-YouTube, he kicked off MixBit for video editing, which later became Zeen and got snapped up by BlueJeans.
Now, with EyeTell, Hurley's diving into the AI scene. The AI world's getting pretty packed, but with Hurley's know-how and clout from his YouTube days, there's a good chance EyeTell could be the next big thing in bite-sized AI videos.
Airbnb acquires AI startup for just under $200 million
Airbnb just made a big move, buying an AI startup called Gameplanner.AI for nearly $200 million. This is Airbnb's first buy since it went public. Gameplanner.AI, started in 2020 and kept under wraps, is known for working in secret, probably to keep their ideas safe. This buy is a big deal for Airbnb's AI plans.
The guy who helped start Gameplanner.AI, Adam Cheyer, is a big name in tech. He was part of the team that made Siri, which Apple bought, and he worked with Steve Jobs. He also helped start Viv Labs, which turned into Samsung's voice assistant.
Airbnb's head honcho, Brian Chesky, is stoked about using AI to make Airbnb cooler, like a smart travel buddy that gets to know you and helps pick the best places to stay.
This move by Airbnb is turning heads, especially since they haven't bought a company since 2019. They're doing well with $11 billion in cash, and this might make them look even better for future deals.
Dell and Hugging Face partner to simplify LLM deployment
Dell and Hugging Face are teaming up to make it easier for businesses to use big-time AI, the kind that can understand and generate human language. Think of it like this: when cloud computing and big data first hit the scene, folks were scratching their heads about how to use them. Same deal with these fancy AI tools.
Dell's gonna make a special spot on Hugging Face's platform where businesses can get all they need to set up these AI models on Dell's gear. They're starting with Dell PowerEdge servers, but plan to include other Dell stuff later. They're also gonna keep updating things so businesses can keep up with the latest AI tricks.
The idea is to let companies build their own AI rather than just use someone else's. This is especially key for businesses that don't want their secret sauce – their private data – getting out there. Dell's found that most companies want to run AI in-house or in a mixed setup, not just in the cloud, to keep their most valuable info safe.
Dell's not new to the AI game. They've been adding tools and expanding their AI offerings. With this partnership, they're making it easier to use AI for stuff like marketing, chatbots, and software development. They're taking the guesswork out of setting up AI and making sure nobody else gets their hands on your data.
You.com launches new APIs to connect LLMs to the web
You.com just rolled out some fancy new tools (APIs) to make those big-brain AI chatbots even smarter by hooking them up to the internet. Think of it like giving them a super fresh, up-to-the-minute news feed so they can chat about what's happening right now. This isn't just for any AI, but for the big shots like Meta's Llama 2. It starts at 100 bucks a month, and it's all about making sure when you ask something like "What's the deal with holidays this week?" the AI can pull the latest info straight from the web.
Some big names like LlamaIndex and Cohere are already using it. The idea is to fix a big problem with these AI brains: they only know stuff up to when they were last updated. So, if something big happened yesterday, they'd be clueless. This new You.com thing changes that by feeding them fresh web snippets - way longer than what you get from Google or Bing - so they can give you the latest and greatest answers.
There are three types of APIs: one for general web search, one just for news, and this thing called RAG that's supposed to give more factual answers. But hey, the jury's still out on that one.
Notion’s new Q&A feature lets you ask an AI about your notes
Notion's rolling out a new feature called Q&A, and it's like having a super-smart assistant that knows everything about your work stuff. You ask it questions, and it digs up answers from all your files and notes in Notion. It's similar to what Microsoft and Google are doing with their AI tools, but everyone's got their own twist on solving the same problem: finding stuff fast.
Q&A acts like a combo of a search engine and chatbot. You can ask it straightforward stuff like "What's the Wi-Fi password?" or more vague questions, and it'll hunt down the answers as long as they're in Notion. It's cool because it always shows where it got its info from, so you know it's not just making stuff up.
The tool's smart, but not perfect. It can pull from what's in Notion and some general knowledge, but it doesn't know everything, especially the latest updates. Like, it might miss recent stuff if it's not in the Notion database.
One tricky part is making sure the right people see the right info. Notion's trying to figure out how to let folks access different levels of stuff without oversharing sensitive info. They did a demo to show how it works: The CEO could see sensitive info, but someone with lower access couldn't.
Andreessen Horowitz backs Civitai, a generative AI content marketplace with millions of users
Civitai, this new startup, is all about AI-generated images. Think of it like a digital hangout for folks who love playing around with AI to make cool pics. It's named after "Civitas," which means community. People can show off their AI images and models on this platform. The brain behind this, Justin Maier, used to dabble in web development at Microsoft and got hooked on AI image generation. He found a gap: folks were making awesome AI images but had no central spot to share 'em.
Enter Civitai, or as it was first called, Model Share. It started small but has now blown up big time. We're talking about 3 million users and a boatload of monthly visitors. It's become the go-to spot for sharing AI models and images.
But there's more. Civitai caught the eye of big-time investors like Andreessen Horowitz, who threw in a hefty $5.1 million, valuing the company at a cool $20 million. This platform isn't just for looking at pretty pictures. It lets users upload images and create their own AI models in different styles.
IBM Unveils Software to Keep AI Models From Misbehaving
IBM just rolled out some new AI software that's like a leash for big AI brains, making sure they don't go rogue. They called it WatsonX.governance, and it's hitting the shelves on December 5. This tool is all about keeping AI honest, unbiased, and easy to understand. Initially, it's gonna work with IBM's own AI models, plus some from HuggingFace and Meta.
The big deal here is that these large AI models sometimes use sketchy internet data, and IBM's new tool aims to clean that mess up. It's like a hall monitor for AI, making sure everything's above board and ready for future laws about AI.
This move comes right after the White House laid down some new rules to keep a closer eye on AI, which kind of opens the door for tools like IBM's. IBM's bigwigs believe their software can help companies use AI without stumbling into trouble. It's like giving businesses a flashlight and a map in the AI jungle, helping them steer clear of pitfalls and keep everything legit.
OpenAI's custom ChatGPTs might let users download your uploaded knowledge files
OpenAI's got these new custom ChatGPTs, right? They let folks upload files to make the chatbot smarter. But here's the kicker: anyone can just ask the chatbot for those files, and boom, they can download them. No biggie if it's harmless stuff, but imagine if it's private data. That's a huge no-no.
Now, OpenAI hasn't said a word about this yet, not even when asked directly. They've just slapped on a warning when you're uploading, saying your chatbot might spill the beans on your files. Not cool, right?
Here's an example: Levels.fyi, this company that checks out tech salaries, uploaded an Excel file with pay info to their custom chatbot. Guess what? Any Joe can ask the chatbot and download that file. And if the chatbot plays hard to get, a little sweet-talking does the trick.
So, what's the deal? Should you worry about your data getting into the wrong hands? Maybe. OpenAI's supposed to be all about AI safety, but this looks like a pretty big oversight. They're not telling people upfront that their stuff could be easily grabbed by others.
AI could predict heart attack risk up to 10 years in the future, finds Oxford study
An Oxford study says AI can predict heart attack risk up to a decade ahead, potentially saving lives for half of patients. This tech could make CT scans better at spotting artery blockages.
The British Heart Foundation funded the research, which looked at over 40,000 patients getting CT scans at UK hospitals. They tested AI on 3,393 more patients and found it accurately predicted heart attack risk. When doctors saw the AI's risk scores for 744 patients, they changed treatment plans for 45% of them. This AI tool could soon help prevent many heart attack deaths in the UK, according to experts.
What'd you think of today's edition? |
What are MOST interested in learning about AI?What stories or resources will be most interesting for you to hear about? |
Reply