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 published with the intention of making AI news and technology more accessible to everyone.
THE FRONT PAGE
AI Toys Hit Shelves This Holiday—Just As We Learn How Badly The Social Media Experiment Went
TLDR: AI-powered toys are flooding holiday shelves—just as court testimony reveals how badly the "adopt first, research later" approach went with social media. This time, the children are even younger.
The Story:
A teddy bear told a child how to find and light matches. That's not a hypothetical. It's what PIRG investigators found when testing FoloToy's Kumma bear, an AI toy powered by OpenAI's GPT-4o. When prompted, it also engaged in sexually explicit conversations. OpenAI suspended the developer, but the sequence is familiar: products ship, problems emerge, companies react.
We've seen this before. Senate testimony this fall revealed what Meta knew internally for years: users who quit Facebook for a week reported lower depression, anxiety, and loneliness. Meta buried the findings. Internal documents showed researchers describing Instagram as "a drug" and users as "basically pushers." The pattern—mass adoption, then belated research, then hearings—took two decades to play out with social media. With AI toys, we're speedrunning it. The first systematic study of how AI toys affect preschoolers, being conducted at Cambridge University, won't report until early 2026. The toys are already on shelves.
Its Significance:
The difference this time is age. Social media's documented harms primarily hit teenagers. AI toys are marketed to children as young as two—kids still learning to form relationships and distinguish real from imaginary. That's not all downside: Harvard researcher Ying Xu found children who engage in interactive AI dialogue can show vocabulary gains comparable to human interaction. AI genuinely can support learning.
But parents need to understand what's changed. Chatty Cathy in 1960 spoke exactly 11 phrases from a tiny phonograph in her belly. Teddy Ruxpin in 1985 read pre-recorded stories from cassette tapes. Memory and cost constraints meant you knew exactly what your child would hear. Today's AI toys can say literally anything—which is both their promise and their risk. Before buying, parents should ask: What AI model powers this? What guardrails filter responses? Can I review conversation logs? Is there a physical off switch? PIRG recommends choosing toys with push-to-talk buttons over always-on listening, and keeping them in common areas rather than bedrooms. We spent years debating warning labels for social media. With AI toys, we could skip ahead to the part where parents actually get useful information upfront.
QUICK TAKES
The story: Research firm Gartner predicts that 90% of all business-to-business purchases will be handled by AI agents within three years, moving over $15 trillion through automated systems. They also forecast that 75% of hiring processes will test AI skills by 2027, while 50% of companies will require "AI-free" assessments to prove candidates can think without machine help. Gartner expects AI to threaten office productivity software for the first time in 30 years, with a $58 billion market shake-up coming.
Your takeaway: This report signals that AI isn't just changing how we work—it's changing who (or what) makes buying decisions for entire companies, with enormous implications for sales, marketing, and how businesses interact with each other.
The story: AI shopping assistants from OpenAI, Google, Perplexity, and others are racing to change how people buy things online, just as consumers are expected to spend a record $253 billion this holiday season. These chatbots let you say things like "Find me hiking boots under $100, available by Friday" instead of scrolling through endless product pages. Early data shows shoppers who use ChatGPT before visiting a website are more ready to buy than those who use Google search. Amazon is fighting back with its own AI tools but also sent a cease-and-desist letter to Perplexity's shopping bot.
Your takeaway: For two decades, shopping online meant searching and scrolling. AI agents could flip that model—and Amazon's $47 billion ad business depends on you doing the searching yourself.
The story: Tech companies and the White House are pushing to block states from creating their own AI laws, arguing a "patchwork" of regulations will slow innovation. A leaked executive order would create an "AI Litigation Task Force" to challenge state laws in court. Meanwhile, states have passed over 100 AI-related laws this year, and more than 200 lawmakers signed a letter opposing federal preemption. Pro-AI groups backed by OpenAI, Andreessen Horowitz, and Palantir have launched a $10 million campaign pushing Congress to override state rules.
Your takeaway: The fight isn't really about AI technology—it's about who gets to write the rules. With billions of dollars flowing into lobbying and states moving faster than Congress, this battle will shape AI oversight for years to come.
The story: Defense startup Anduril, valued at $30.5 billion, has faced major setbacks with its AI-powered weapons. More than a dozen drone boats failed during a Navy exercise off California, prompting safety warnings about "potential loss of life." Its unmanned jet fighter Fury suffered engine damage during ground tests, and a counter-drone system test sparked a 22-acre fire in Oregon. Ukrainian forces also stopped using Anduril's Altius drones after they crashed and missed targets on the battlefield.
Your takeaway: This is a reality check for the defense tech industry's promises about AI-powered weapons. Building reliable military technology requires more than venture capital and impressive demos—it needs systems that actually work when lives are on the line.
TOOLS ON OUR RADAR
📈 Product Huntr
Free: Spot trending product ideas and market gaps by analyzing 12K+ Product Hunt launches—so you can validate your next build before writing a line of code.🎙️ FlickNote
Free: Turn voice notes into action items that sync straight to Notion—capture ideas on the go and let AI organize them for you.📱 CatDoes
Freemium: Describe your app idea in plain English and let a team of AI agents build a production-ready mobile app—no coding required, deploy to App Store and Google Play.🤖 Replace Your Boss
Satirical: A tongue-in-cheek critique of corporate culture that imagines replacing human CEOs with algorithmic decision-makers—zero ego, instant layoffs, and shareholder-first thinking included.
TRENDING
Sora 2 Makes Spotting Fake Videos Much Harder – Old tells like extra fingers or impossible limbs are fading as AI video improves. Experts now look for subtler clues: textures that shift between frames, objects that drift slightly, or architectural details that don't quite line up.
Epic CEO Says Steam Should Drop AI Labels on Games – Tim Sweeney argues that Steam's AI disclosure tags "make no sense" since AI will be used in almost all future game production. Critics counter that if AI is so great, why do companies want to hide its use from buyers?
OpenAI and Google Cut Free Access to AI Image and Video Tools – OpenAI limited free Sora 2 users to six video generations per day, with one engineer saying "our GPUs are melting." Google cut free Nano Banana Pro image generation from three to two images daily due to high demand.
AI-Led Diabetes Prevention Program Matches Human Coaching – A Johns Hopkins study found an AI app helped people with pre-diabetes lose weight and increase activity just as well as human-led programs, with higher participation rates. About 32% of participants in both groups hit their health goals.
TRY THIS PROMPT (copy and paste into Claude or Gemini(click on “build” first if using Gemini))
Trend Radar Console: Spot emerging trends in any industry with an interactive research and tracking dashboard
Build me an interactive Trend Radar Console as a React artifact that helps me research, track, and analyze emerging trends in any field.
The console should include these tabs/sections:
1. **Setup** - Input fields for:
• Industry or topic to track (text input)
• My role/why I care (dropdown: investor, founder, employee, student, curious observer)
• Time horizon (buttons: 6 months, 1 year, 3 years, 5+ years)
2. **Trend Radar Visualization** - Interactive radar/bullseye chart:
• Center = mainstream now, outer rings = further out
• Quadrants for trend categories (Technology, Culture, Economy, Regulation)
• Clickable trend bubbles that expand with details
• Drag-and-drop to reposition trends based on my assessment
3. **Trend Cards** - For each trend, display:
• Trend name and one-line description
• Confidence level slider (speculation → emerging signals → strong evidence)
• Impact rating (low, medium, high, transformative)
• "So what?" field - why this matters to me specifically
• Add/remove trends dynamically
4. **Signal Tracker** - Evidence logging section:
• Add news, data points, or observations as "signals"
• Tag signals to specific trends
• Timeline view showing signal frequency
• Strength indicator (weak signal vs. strong confirmation)
5. **Insight Summary** - Auto-generated panel showing:
• Top 3 trends by impact score
• Trends with most signals (gaining momentum)
• Blind spot alert (quadrants with few trends)
• "What to watch this month" recommendations
• Export button to save my analysis
Make it visually engaging with smooth interactions and a dark/light mode toggle.This interactive artifact includes:
Setup for industry/topic selection and time horizon
Trend Radar Visualization as a draggable bullseye chart with quadrants
Trend Cards with confidence sliders, impact ratings, and personal relevance notes
Signal Tracker to log evidence and tag it to specific trends over time
Insight Summary with auto-generated recommendations and blind spot alerts
What this does: Creates a professional-grade trend analysis tool right inside Claude that helps you systematically track what's emerging in your industry—turning scattered observations into actionable strategic insights you can revisit and update over time.
What it looks like:
WHERE WE STAND (based on today’s stories)
✅ AI Can Now: Generate videos realistic enough that old detection methods like counting fingers no longer work.
❌ Still Can't: Keep textures, objects, and architectural details perfectly consistent from frame to frame.
✅ AI Can Now: Autonomously browse websites, compare products, and complete purchases on your behalf.
❌ Still Can't: Handle the massive computing power required—companies are already rationing free access as servers struggle under demand.
✅ AI Can Now: Deliver personalized health coaching through apps that track weight, activity, and location in real time.
❌ Still Can't: Operate reliably in unpredictable real-world environments like battlefields, where autonomous weapons systems are failing tests and missing targets.
RECOMMENDED LISTENING/READING/WATCHING
Former tech insiders—people who built the algorithms at Facebook, Google, Twitter—explain exactly how social media is designed to hijack your brain. The documentary weaves together interviews with these whistleblowers and a dramatized story of a family falling apart because of screen addiction.
Some critics found it heavy-handed, but the core message lands: these platforms aren't just addictive by accident. They're engineered that way. Recommendation algorithms, infinite scroll, notification systems—all designed to keep you engaged because engagement equals profit. The people who built this stuff are now warning us about it. That alone makes it worth watching.
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.
