🧟 That AI Stat? It Died Years Ago

The web's zombie stats end up in AI answers, NASA trains a robot medic, a supercomputer turns up on eBay, and more...  ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌

October 30, 2026   |   Read online   Hey, Jhon! Welcome back to the #1 AI newsletter in the world! Here’s what we have today: Zombie stats keep showing up in AI answers. NASA is training a robot medic for Mars. AI programs from 1961 and 1984 run again. Someone listed a $60M AI supercomputer on eBay.       Fact-Checking Zombie Stats Keep Coming Back Source: Trakkr There’s a number that’s been bouncing around the internet for years, that 75% of people never click past the first page of Google. And it’s easy to believe. Page two of Google might as well be a ghost town. Ask an AI about it today and you may get the same number back. But a team that went looking for the study behind it only found it credited to an old marketing roundup that doesn’t even carry it anymore. It turns out a lot of numbers online are like that. They call them zombie stats, numbers that died at the source a long time ago (or never had one) and keep walking around anyway. To see how many are out there, the team asked a bunch of popular AIs questions that need a number for an answer, then followed the stats back through the links. Fewer than half led back to a study that said the same thing. Even the one about websites, that 94% of first impressions are about design, started out a lot smaller. Back in 2004, it was a study where 15 people looked at health websites, and nearly all of their reasons for turning a site down were about the design. Twenty years of retelling later, it’s a rule for every website out there, and that’s the version AI gives you. To be fair, none of this started with AI. These numbers lived in blog posts and slide decks long before chatbots, and AI learned from that same web. As for page two of Google, the team still couldn’t find a single study behind it. Our take: A specific number feels like it has to come from somewhere, and that’s exactly how these stick around. Before you use a stat for something that matters, ask the AI for the original study and give it a quick look. If you can’t find who was asked and when, treat it like a rumor.       Together with Wispr Flow Messy Thoughts Become Clean Prompts The full version of what you want from AI starts out as a ramble in your head. This week’s dinners, plus the kid who won’t touch anything green. By the time you’ve typed it into a tidy request, half of those details didn’t make the cut. Wispr Flow lets you say it the messy way and hands your AI a clean, structured prompt instead. Talk it through out loud, tangents and all, and Prompt Engineer sorts it into a prompt with the task and the details laid out where the AI can’t miss them. You can set it to run after every dictation, which means each ramble comes out as a prompt without you doing anything extra. It works in most apps on your computer where you type, chatbots included.       Space Medicine NASA's Robot Medic Practices for Mars Source: Google Cloud Somewhere on the way to Mars, an astronaut wakes up with a sharp pain in their back. On the space station, a doctor on the ground could talk the crew through it live. Out near Mars, a message home and the reply can take up to 44 minutes, which is a long time to wait when something hurts. NASA and Google are working on a robot that can step in as the crew’s extra pair of hands. In that back-pain moment, it’d grab the ultrasound machine and get it ready, while the crew’s medical officer focuses on the patient. The rest of the time, it’s more of a helper around the ship, hauling cargo and keeping tools where they belong. Nobody has to program its moves one by one, either. The crew types out what they need in everyday words, and an open-source AI model turns those words into motion. It practices in a simulated world first, and since there’s no reliable internet that far out, everything runs on the robot itself. In an early demo at NASA’s Johnson Space Center, it found a medical bag and moved it where it needed to go. Early lab tests had it getting that kind of task right a little more than eight times out of ten. The medical officer still makes every call. The robot’s job is fetching and setting up, which leaves the person treating the patient with both hands free. Our take: Space has a habit of sending its tech back home. If a robot can set up an ultrasound for a crew that far from help, the same trick fits pretty naturally in a rural clinic or the back of an ambulance (anywhere a second pair of hands is hard to come by).       Coming in Hot AI Tools of the Day 📊 AnswerScout puts your buyers' questions to five AI assistants daily, then scores how often you come up and which rivals land ahead of you. 🛎️ Sundae handles the errands you text it, like booking a dinner table or comparing flights, in a browser it runs itself and checks with you before paying. 🔎 AEO Action drafts the pages that get your business named when someone asks AI for a recommendation, and holds every one until you approve it. 🎬 pitchboom.io turns a link to whatever you sell into a finished video ad with its own script and voice-over, then translates it into 34 languages. 🤖 Impeccable gives your AI coding agent design commands like critique and polish, plus a checker for cookie-cutter layouts, free and open source on GitHub.       AI History Two Old AI Programs Wake Up Source: Jeff Shrager and team Herbert Simon won a Nobel Prize and helped start the field of AI. In 1961, he wrote a program that tried to write programs the way people do. It compared what it had with what it needed, picked a step that closed the gap, and kept going until the program was done. Then in 1984, a team at MIT built something that sounds a lot like the coding assistants people use today. You’d type commands in a stiff, limited kind of English, and it’d snap together familiar chunks of code that good programmers reuse all the time. Both programs ended up sitting on a shelf for decades. Now a small team has brought them back to life, and they have a name for it, executable archaeology. Instead of reading about old software, you get it running again and watch what it does. Much of the hands-on rebuilding was done by Claude Code, one of today’s AI coding tools. They did that on purpose. Both old programs were attempts to figure out how programmers think, and now a programming AI was helping dig them up. It worked. The 1961 program turned out the same code Simon printed in his 1963 paper, and the 1984 one ran through the same square-root example from its own self-test that December. Simon’s program got to run once more, 65 years after he wrote it, with an AI doing much of the work. Our take: Restoring a classic car teaches you more about how cars work than any manual, and running old code does the same for software. A lot of what feels brand new in AI has roots in ideas people were testing before most of us owned a computer.       Hardware AI Supercomputer Lands on eBay Source: Dexerto Somebody put a $60 million AI supercomputer on eBay. The listing was 100% real, and it even had a Make Offer button. A hardware reseller in Minnesota listed 300 racks of computers, and they said they came out of a big cloud provider’s data center. Each rack stands about as tall as a refrigerator and is packed with Nvidia’s A100 chips, the workhorse behind a lot of AI a few years back. Add it all up, and that’s 9,600 of those chips. The seller also said it all still worked. The fine print reads like any other eBay listing. No returns, and you can pick it up yourself in person if you want. The listing has since ended, but there were two of these up for grabs. Our take: Keep an eye on where hardware like this ends up. Big AI companies keep moving to newer chips, and the older ones still have plenty left in them. That’s how universities and smaller labs get their hands on computing power that used to be out of reach.       Open Source Give Your Files an AI Tutor Source: University of Hong Kong A textbook will happily let you read chapter 3 four times without ever checking whether any of it stuck. DeepTutor is a free, open-source AI tutor built to fix that. And it works with whatever you’re studying. Drop in a textbook or a PDF, and you can ask it questions in your own words. Each answer points back to the part of your material it came from. You’re never left guessing where it got something. From there, it gives you a few ways to make sure it sticks. Quizzes and a daily practice round built from your own material. A progress page that shows what you’ve got down and what still needs work. YouTube lectures you can search, with help tied to the exact moment in the video. An interactive book with flash cards, made from your sources. You also pick the AI behind it. Connect one you already use, or run a free one on your own computer through Ollama, and your notes never leave your desk. It picked up more than 40,000 stars on GitHub (the site’s version of a bookmark) in its first nine months, and it’s still getting regular updates. The catch: There’s no regular app to download yet. Getting it running means installing Docker, a free program for running apps like this, then pasting in one command from the project page. It takes a few minutes the first time, and after that it opens in your browser like any website. Our take: It’s fair to wonder why you’d bother when any chatbot will quiz you on request. The difference is that everything stays tied to your own material and your own progress, so each study session picks up where the last one left off instead of starting from zero.       Prompt of the Day Explain It in Five Levels Important Note: You’ll need to click the button to get the complete prompt. View and copy today’s full prompt This prompt has you explain something you know well five times, to a child, a teenager, a college student, a grad student and an expert, all played by the AI. You finish with your weakest explanation rebuilt and tested again.       That’s all for Friday! Before you go, turn your next ramble into a prompt with Wispr Flow. Talk through whatever you’ve been meaning to ask AI, tangents and all, and see what lands in the chat box. How did we do today? Rate this issue, then tell us what you thought. We value your input and will use it to make TAAFT better for you. ⭐⭐⭐⭐⭐ Absolutely Loved It ⭐⭐⭐⭐ Better Than Usual ⭐⭐⭐ Meh, It Was Okay ⭐⭐ Could Be Better Thanks for reading, — There’s An AI For That     Reach out to over 1.8 million AI enthusiasts today. Become a sponsor Update your email preferences or unsubscribe here © 2026 There's An AI For That® SRL 228 Park Ave S, #29976, New York, New York 10003, United States