Beginners in AI
Good morning and thank you for joining us again!
Welcome to this daily edition of Beginners in AI, where we explore the latest trends, tools, and news in the world of AI and the tech that surrounds it. Like all editions, this is human curated and edited, and published with the intention of making AI news and technology more accessible to everyone.
*Over the next few days I'm using the time to migrate to a new hosting platform in an effort to deliver a better experience for my readers. If you notice any delay or hiccup receiving an issue, I apologize in advance, and I thank you for reading along with me. If an issue doesn't show up at its usual time, check your spam or promotions folder. Replying to the email, or marking it "not spam" or "important," helps your inbox provider learn to deliver the email straight to your primary folder going forward.
THE FRONT PAGE
AI Vehicles Went to War for 9 Months, and Soldiers Still Had to Drive
TLDR: A U.S. company has run more than 100 self-driving vehicles in Ukraine's war for nine months, and it's one piece of a fast shift toward robots fighting on the ground, in the air, and soon maybe on two legs.
The Story:
Forterra, a Maryland defense company, told TechCrunch that its self-driving ATVs have run more than 1,100 missions across Ukraine since last October. The vehicles, called Lancers, are built on Polaris ATVs and haul 750 kilograms, about three times what Ukraine's battery-powered robots carry. Over nine months they drove 2,500 miles, moved 777,440 pounds of supplies, and pulled 52 wounded soldiers out of danger. Some got stuck in deep mud and were destroyed by Russian attacks. This is just one piece of a bigger change: Ukraine plans to buy around 30,000 ground robots this year, up from about 2,000 in 2024. China has been showing off its own robot dogs that carry rifles and anti-tank missiles, and a U.S. startup called Foundation wants to build 50,000 humanoid war robots by the end of 2027.
Its Significance:
War is changing fast, and machines are taking on jobs that used to get people killed. Ukraine's military says robots have cut its casualties by as much as 30 percent by keeping soldiers out of the deadliest spots. But one thing gets left out of the hype. These "self-driving" vehicles mostly aren't driving themselves yet. Soldiers steer them by remote control because the AI still can't spot a surprise enemy and react like a person can. So the real story isn't robots replacing soldiers. It's robots doing the deadliest grunt work while humans stay in charge of the driving, at least for now.
QUICK TAKES
The story: Microsoft has started running tens of thousands of Excel and Outlook AI requests each week on its own cheaper MAI models instead of ones from OpenAI and Anthropic, Bloomberg reported. Microsoft's AI chief said the goal is to cut and eventually stop paying Anthropic, which he called "extremely expensive."
Your takeaway: The top AI models cost a lot to run, and even giants like Microsoft are hunting for cheaper options. That could mean you end up using a weaker model inside familiar apps without knowing it.
The story: The U.S. Commerce Department cleared OpenAI to release its most advanced model, GPT-5.6, to everyone starting Thursday. Last month the government made OpenAI limit access to a few approved partners while it tested the model for safety risks.
Your takeaway: The government is now checking powerful AI models before they go wide, kind of like a safety inspection. Anthropic went through the same thing in June with its Fable and Mythos models.
The story: A University of Manchester researcher argues that colleges are too focused on catching AI cheating and not focused enough on teaching students to work alongside AI. The study says human skills like judgment, communication, and adapting to change will matter more as automation spreads.
Your takeaway: If you're in school or paying for someone who is, this is worth thinking about. The classes that teach you to use AI well may end up mattering more than the ones that ban it.
TOOLS ON OUR RADAR
🎧 Libation Free and Open Source: A remarkably simple desktop application that allows you to safely download organize and backup your entire Audible audiobook collection so you permanently own the media you purchased.
💬 Ferdium Free and Open Source: A desktop application that combines all of your favorite messaging services like WhatsApp Telegram and Discord into one single organized window to prevent constant tab switching.
🔐 Element Free and Open Source: A secure communication platform that provides encrypted private messaging and group voice calls without relying on massive technology companies to host your personal conversations.
💽 WinDirStat Free and Open Source: A visual utility that scans your computer hard drive and generates colorful interactive blocks to show you exactly which hidden files are taking up all of your storage space.
TRENDING
A Complete Coding Beginner Built a Military App Using Just AI Chatbots - An Air Force cadet with no coding background used Claude, ChatGPT, and Gemini to "vibe-code" a working battlefield software tool over three months. MIT says it shows non-techies can now build real apps for their own problems.
Meta's New Muse Image AI Can Pull Real Instagram Users Into Fake Photos - Meta launched an image generator that lets people tag any public Instagram account and drop that person into AI-made pictures. It's turned on by default, so you have to dig into your settings to opt out.
Anthropic's Claude Cowork AI Assistant Comes to Phones and the Web - You can now hand Claude a task at your desk and check on it from your phone while it keeps working in the background. Anthropic's own data shows most people use it for business and writing work, not coding.
Musk Says SpaceXAI Will Release Grok 4.5 to Everyone This Week - Elon Musk claims his new Grok 4.5 matches Anthropic's top Claude model but runs faster and costs less. It's built on a 1.5 trillion-parameter system and lands the same week as OpenAI's GPT-5.6.
Discord Says an AI Bug Wrongly Banned Over 8,000 People for Harmless Images - Discord's automatic safety system flagged things like spreadsheets, chessboards, and game textures as illegal content and banned the accounts. Everyone affected is being let back in.
AI Is Turning Into a Bargain Hunter's Market, With a Few Pricey Models on Top - The cost of GPT-4-level AI has dropped about 55 times in under four years, while top frontier models keep getting pricier. About 75 percent of companies now mix cheap and expensive models depending on the job.
TRY THIS PROMPT (copy and paste into Claude, ChatGPT, or Gemini)
🍳 List what's actually in your kitchen. Get a real meal you can make right now, no shopping trip required.
Build a single-file HTML app with vanilla HTML/CSS/JS. The Fridge Forager — list what's in your kitchen, get a real meal you can make right now, no shopping. Persist to localStorage key 'fridge_forager_v1'.
Aesthetic: warm dark kitchen (#17130f), terracotta (#d98a58) primary with warm glow top-left, sage (#8a9a5b) secondary with glow bottom-right. Bitter serif (italic for prompts/teaser/tip) for headings, Manrope sans for body, JetBrains Mono for labels. Sage pill tags for "using what you've got," terracotta for meta chips and steps.
Form: kitchen-contents textarea (encourage a raw unedited list), time dropdown (15 min / 30 min / 45+ min), people-fed dropdown (just me / +1 / small family), dietary dropdown (none/vegetarian/vegan/gluten-free/low-carb), optional mood/craving text input.
System instructions: resourceful encouraging home cook making real meals from what's on hand, no grocery run; assumes only common pantry basics (salt, pepper, oil, maybe butter/flour) unless more is listed; gives ONE confident main recommendation plus two smaller backups, not an overwhelming list; respects dietary restrictions strictly; honest if something's a stretch; genuinely appetizing, not sad leftover food. Return raw JSON: meal_name, teaser (1-2 sentences), meta_chips (3-4 tags), uses (ingredients from their list used), might_need (optional note on 1-2 common extras, or empty string), steps (4-6, each with text + optional time), tip (one practical cook's tip), alternates (2 backup ideas: name + 1-sentence desc).
Render: gradient meal-name hero card with teaser + meta chips. Sage "using what you've got" pill-tag card. Optional terracotta "if you happen to have it" note. Numbered step timeline connected by a gradient line, each with step text + optional time. Cook's-tip card. Two backup-idea mini cards below a "if that's not landing" header. Copy recipe + archive of past meals by name.What this does: List whatever's in your fridge, freezer, and pantry exactly as is, then set your time, who you're feeding, dietary restrictions, and optional mood or craving. It hands back one confident meal built almost entirely from what you listed (assuming only basic staples like salt, oil, and pepper), with a teaser, quick meta tags, a visual list of the ingredients it's using from your kitchen, a note if it needs one or two common extras, clear numbered steps with rough timing, and a cook's tip. If the main idea doesn't land, two simpler backup ideas sit underneath. Saves every foraged meal to localStorage.
What this looks like:
WHERE WE STAND(based on today’s news)
✅ AI Can Now: Drive supply and rescue vehicles across rough battlefield terrain on its own (Forterra's Lancers in Ukraine).
❌ Still Can't: Spot a surprise enemy and react in real time, which is why soldiers still steer by remote control.
✅ AI Can Now: Help a total coding beginner build a working software app from plain prompts (MIT vibe-coding case).
❌ Still Can't: Reliably tell a harmless chessboard or spreadsheet apart from banned content (Discord's 8,000 wrong bans).
FROM THE WEB
RECOMMENDED LISTENING/READING/WATCHING
Serial Experiments Lain (1998) - Anime TV Series
A 13-episode Japanese anime that came out just before The Matrix, about a shy schoolgirl named Lain who receives an email from a classmate who supposedly died the week before, and starts spending more and more time on a global network called the Wired that is starting to bleed into the real world. Directed by Ryutaro Nakamura with character design by Yoshitoshi ABe, and you’ll see how much its style influenced the cyberpunk that came after it.
Thank you for reading. We’re all beginners in something. With that in mind, your questions and feedback are always welcome and I read every single email!
-James
By the way, this is the link if you liked the content and want to share with a friend.
Some * designated product links may be affiliate or referral links. As an Amazon Associate, I earn from qualifying purchases. This helps support the newsletter at no extra cost to you and Amazon makes a tiny hair less.
