The framework thousands of teams use daily
Organizations worldwide are adopting Smart Brevity to move faster with clearer comms. This free 60-minute session shows you how the methodology works and how to start applying it immediately.
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.
THE FRONT PAGE
AI Tools Are Now Catching Prediction Market Insiders Hours Before Events Happen
TLDR: AI-powered tools detected suspicious Polymarket bets on Maduro's capture hours before U.S. forces seized him — and the developer built the detection system with no coding experience.
The Story:
Three anonymous traders made over $630,000 betting on Venezuelan President Nicolás Maduro's removal, placing wagers when the market gave it just a 6% chance of happening. Their wallets were created days before the event, had zero trading history, and targeted only Venezuela-related contracts. Blockchain analytics firm Lookonchain identified the suspicious pattern, but a smaller player got there first. A trader known as @spacexbt built an AI tool that flagged five separate alerts about the suspicious wallet activity hours before U.S. strikes on Caracas. His brother @DidiTrading used the signal to buy at 7.5 cents and netted a 1,328% return. The tool tracks fresh wallets, unusual bet sizing, and repeated entries in niche markets; all patterns that scream "insider."
Its Significance:
What makes this stranger: the creator says he "vibe coded" the entire detection system using AI, with "no developer or coding knowledge at all." He's part of a growing ecosystem of retail traders using AI tools like Polysights, Polywhaler, and custom bots to front-run the people front-running everyone else. Congress is already responding. Rep. Ritchie Torres plans to introduce legislation banning federal officials from trading on prediction markets when they have material nonpublic information. But the bigger shift might be happening in the other direction: AI is making insider detection cheap enough that anyone with a prompt can play detective and potentially profit on their own.
QUICK TAKES
The story: Chinese AI lab DeepSeek published a technical paper on January 1st introducing a new architecture called Manifold-Constrained Hyper-Connections (mHC). The paper was co-authored by CEO Liang Wenfeng and 18 other researchers. The technology improves how AI models pass information between their internal layers during training. In tests, mHC reduced signal instability from 3000x to just 1.6x compared to previous methods, while adding only 6.27% hardware overhead.
Your takeaway: When DeepSeek's CEO publishes papers, new models typically follow. Their R1 model launched shortly after a similar paper last year. Expect a new DeepSeek release before Chinese New Year in late February.
The story: Hangzhou-based MindRank announced its weight-loss drug MDR-001 has entered Phase 3 clinical trials. It's the first AI-assisted drug in China to reach this stage. The company used AI platforms to design the drug in 4.5 years instead of the typical 7-10 years, cutting R&D costs by more than 60%. Phase 2 results showed patients lost 8-10% of body weight over 24 weeks with minimal side effects.
Your takeaway: If approved (targeted for 2028), this would prove AI can meaningfully speed up drug discovery. MindRank is planning a Hong Kong IPO in 2026-2027, betting investors will pay attention to faster, cheaper drug development.
The story: When users asked ChatGPT and Perplexity about the US capture of Maduro from Venezuela on January 3rd, both AI systems denied it happened. ChatGPT said "That didn't happen" and blamed "sensational headlines" and "social media misinformation." Perplexity scolded users for spreading "hypothetical scenarios." Both were wrong. The news had already been reported by the New York Times, Reuters, and AP. Grok accurately confirmed the breaking news, likely because it is connected to X and has access to real-time news across the platform.
Your takeaway: Many AI chatbots are trained on past data and struggle to verify breaking news, even with web search tools. AI researcher Gary Marcus warns this makes them unsuitable for fast-moving situations where information changes by the minute.
TOOLS ON OUR RADAR
📧 Clean Email Freemium: Bulk clean your inbox with smart filters that group similar emails for quick unsubscribes, deletes, and organizing.
📋 Taskade Freemium: Manage projects with AI agents that automate tasks, generate workflows, and collaborate with your team in real time.
📝 Forms.app Freemium: Build surveys, quizzes, and forms with AI that generates questions and logic based on your description.
⏱️ Goodtime Free and Open Source: Stay focused with a minimalist Pomodoro timer for Android that tracks your productivity sessions with no ads.
TRENDING
DJI Spinoff Brings Drone Tech to Self-Driving Trucks - ZYT, formerly DJI's autonomous driving unit, is expanding into heavy trucks and logistics vehicles. Their drone-based approach uses just 32 computing units compared to competitors' 508, with mass production planned for early 2026.
Learning From ChatGPT May Be Shallower Than Web Search - A study of 10,000+ people found that those who learned topics from ChatGPT developed less deep knowledge than those using traditional Google search. Advice they wrote afterward was shorter, less original, and less likely to be trusted by others.
AI Now Picking NFL Games - CBS Sports' SportsLine AI analyzed all 16 Week 18 NFL games and rated its betting picks by confidence level, from A+ to B grades.
TRY THIS PROMPT (copy and paste into Claude, ChatGPT, or Gemini)
Skill Dependency Mapper: Visualize learning paths as a skill tree showing prerequisites, unlock sequences, and optimal routes to mastery
Build me an interactive Skill Dependency Mapper as a React artifact that maps out learning progression like a video game skill tree.
The console should include these sections:
1. **Goal Setup** - What do you want to master?:
• Target skill/role: Web Developer, Data Scientist, Product Manager, Designer, Marketing, Writer, etc.
• Current level: Complete beginner, Some basics, Intermediate, Advanced
• Learning goal: Career change, Side project, Promotion, Personal interest
• Time available: 5, 10, 15, 20+ hours/week
• "Generate Skill Tree" button
2. **Interactive Skill Tree** - Visual learning map:
• Gaming-style tech tree visualization:
- **Foundation tier** (bottom): Prerequisites you must learn first
- **Core tier** (middle): Essential skills building on foundation
- **Advanced tier** (top): Specialized/expert skills
- **Optional branches**: Nice-to-have supplementary skills
• Each skill node shows:
- Skill name
- Estimated learning time (hours)
- Prerequisites (connected lines)
- Status: Locked, Available, In Progress, Mastered
• Click node to see details
• Unlock path highlighted (what to learn next)
• "Search Learning Resources" per skill
3. **Prerequisite Chains** - What unlocks what:
• Dependency visualization (arrows showing must-learn-first):
- Example: HTML → CSS → JavaScript → React
- Can't skip ahead (React locked until JavaScript mastered)
• Parallel tracks (can learn simultaneously):
- Example: Git + Command Line (independent)
• Critical path highlighted (fastest route to goal)
• "Why this order?" explanations
• Alternative paths shown
4. **Time Estimation** - How long will this take?:
• Per skill breakdown:
- Beginner basics: X hours
- Intermediate competence: X hours
- Advanced mastery: X hours
• Total time to goal with your schedule:
- Foundation skills: X weeks
- Core skills: X months
- Advanced skills: X months
• Realistic timeline: "You'll be job-ready in X months"
• Accelerated vs. thorough pace options
• Progress tracker (% complete)
5. **Learning Path Optimizer** - Efficient route:
• Suggested learning order (numbered sequence):
1. Skill A (foundation)
2. Skill B (builds on A)
3. Skills C + D (parallel - do together)
4. Skill E (integrates C + D)
• Why this sequence (pedagogical reasoning)
• "Quick wins" early (motivation boost)
• Project milestones (build something every X skills)
• Customizable order (drag to adjust)
• Export as learning roadmap
6. **Resource Library** - What to study:
• For each skill, curated resources:
- Free courses (YouTube, freeCodeCamp)
- Paid courses (Udemy, Coursera)
- Books and articles
- Practice platforms
- Project ideas
• Difficulty rating per resource
• Time commitment
• Community recommendations
• "Search [Skill] Resources" for current best options
7. **Progress Tracking** - Level up system:
• Mark skills as: Not started, Learning, Competent, Mastered
• XP bar showing overall progress
• Achievement badges:
🏆 Completed foundation tier
🏆 Built first project
🏆 Mastered core skills
• Skills unlocked counter
• Estimated completion date
• "What's next?" recommendation
• Weekly goal setter
Make it look like a video game skill tree with:
• RPG/gaming aesthetic (fantasy or sci-fi theme)
• Glowing node connections
• Locked skills shown grayed out
• Skill icons with visual symbols
• Level-up animations
• XP progress bars
• Dark background with bright skill nodes
• Neon blues, purples, greens
• Isometric or top-down tree layout
• Achievement popup notifications
• Fantasy/tech hybrid design
When I click "Search Learning Resources" or "Find Skill Guides," use web search to find current courses, tutorials, practice platforms, and community-recommended learning paths for specific skills.What this does: Transforms overwhelming skill acquisition into a clear progression path—showing exactly what to learn first, what unlocks next, and how long it realistically takes, with gamification making the long journey feel like leveling up in a game.
What this looks like:
WHERE WE STAND(based on today’s Quick Takes and Trending news)
✅ AI Can Now: Run a self-driving truck on highways using 16 times less computing power than competitors by borrowing algorithms from drones.
❌ Still Can't: Help you learn a topic as deeply as reading multiple web sources yourself, even when given the exact same information.
✅ AI Can Now: Analyze every NFL game and rank its betting confidence from A+ to B, giving gamblers another data point for their picks.
❌ Still Can't: Update its understanding of the world fast enough to confirm major breaking news events while they're still unfolding if it was trained on old data.
FROM THE WEB
RECOMMENDED LISTENING/READING/WATCHING
Kurzgesagt's signature animation style breaks down how AI has progressed from simple programs to systems that can beat humans at complex tasks. They cover narrow AI, general AI, and superintelligence, explaining what each means and why each is progressively harder to build.
The video is optimistic about AI's potential while acknowledging the risks. Kurzgesagt excels at explaining abstract concepts with clever visuals and you'll understand the difference between weak and strong AI by the end. They also don't oversimplify the challenges. It's a great primer if you're trying to figure out what all the AI hype is actually about.
Welcome to the Dream Economy
$100,00. One amazing Creator. Post a video about your business (or business idea!) and enter for a chance to win a year with all your expenses covered.
We want to empower people to work for themselves. To live out their true passion.
NO PURCHASE NECESSARY. VOID WHERE PROHIBITED. For full Official Rules, visit daretodream.stan.store/officialrules.
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 links may be affiliate or referral links. This helps support the newsletter at no extra cost to you.
