In partnership with
Payroll errors cost more than you think
While many businesses are solving problems at lightspeed, their payroll systems seem to stay stuck in the past. Deel's free Payroll Toolkit shows you what's actually changing in payroll this year, which problems hit first, and how to fix them before they cost you. Because new compliance rules, AI automation, and multi-country remote teams are all colliding at once.
Check out the free Deel Payroll Toolkit today and get a step-by-step roadmap to modernize operations, reduce manual work, and build a payroll strategy that scales with confidence.
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 Can Now Clone Anyone's Full Body From a Single Photo for a Few Dollars
TLDR: AI can now swap your entire body, not just your face, and 14 million people just watched it happen to Stranger Things actors.
The Story:
A Brazilian creator named Eder Xavier posted videos this week showing himself transformed into Millie Bobby Brown, David Harbour, and Finn Wolfhard. The clips matched their faces, bodies, movements, and clothing with striking accuracy. Made with Kling AI's Motion Control tool, they racked up over 14 million views on X and caught the attention of a16z partner Justine Moore, who noted that Hollywood's production pipelines will change faster than expected. What separates this from earlier deepfakes is that the entire body gets replaced, not just the face. A similar clip showing a real-time Leonardo DiCaprio swap from Wolf of Wall Street hit a million views the same day.
Its Significance:
Earlier deepfakes left telltale signs. Faces didn't quite match body movements. Visible edges appeared where the swap happened. Full-body swapping removes those detection cues. Security researchers at GetReal told Decrypt that for a few dollars, anyone can now generate full-body videos of a politician, celebrity, CEO, or private individual using a single image. That raises questions about impersonation, fraud, and disinformation as these tools become more accessible. Binghamton University professor Yu Chen noted that detection methods built for face-only manipulation may not work against full-body swaps yet.
QUICK TAKES
The story: Taiwan and the U.S. struck a trade deal that cuts tariffs to 15% and commits $500 billion in Taiwanese investment to build semiconductor and AI production capacity in America. Commerce Secretary Lutnick called it the largest semiconductor investment in U.S. history.
Your takeaway: This deal reshapes the global chip supply chain, with the U.S. aiming to bring 40% of Taiwan's production stateside by 2036. It strengthens U.S. access to advanced AI chips while creating a deeper economic bond between the two countries.
The story: OpenAI invested the largest check in Merge Labs' seed round at an $850 million valuation. The startup, co-founded by Sam Altman, is developing non-invasive brain-computer interfaces that use ultrasound and molecular engineering instead of surgical implants.
Your takeaway: OpenAI is betting that controlling AI with your thoughts will be the next major interface leap. The deal raises questions about Altman's overlapping roles, but signals big tech's belief that BCIs could become as common as smartphones.
The story: Anthropic's latest Economic Index report analyzed 1 million Claude conversations and found AI succeeds on most tasks but struggles with complex ones. The study also found AI tends to handle higher-skill tasks, which could lead to "deskilling" of some jobs.
Your takeaway: This is the most detailed look yet at real-world AI usage patterns. The finding that AI covers more educated tasks suggests white-collar workers may see their jobs change faster than expected, though bottleneck tasks that AI can't handle may protect some roles.
TOOLS ON OUR RADAR
🏠 Home Assistant Free and Open Source: Control your smart home locally with privacy-first automation that supports 2,000+ device integrations.
💼 Bigin by Zoho Freemium: Get a CRM built from scratch for small businesses—not a stripped-down enterprise tool.
🎵 Focus@Will Paid: Listen to neuroscience-based music designed to help you concentrate for hours without distraction.
📚 Readwise Paid: Remember what you read by syncing highlights from Kindle, articles, and PDFs straight into your favorite note-taking app.
TRENDING
CATL Deploys Humanoid Robots on Battery Assembly Lines - The world's largest battery maker became the first to use humanoid robots at scale in production. The robots plug in high-voltage connectors with 99% accuracy and work three times faster than humans by skipping breaks.
AMA Awards $12 Million to Bring AI to Medical Training - The American Medical Association gave grants to 11 institutions to personalize doctor training using AI and big data. The projects will help medical students and residents get individualized learning paths throughout their careers.
BlackRock-Microsoft AI Fund Raises $12.5 Billion - The AI infrastructure venture backed by BlackRock and Microsoft raised $12.5 billion toward its $30 billion target. With leverage, the fund could support up to $100 billion in data center and power infrastructure investments.
Lockheed Martin CIO Says AI Is Reshaping Her Role - The defense contractor's chief information officer says AI is centralizing technology decisions across the company. The change signals broader shifts in how large corporations manage tech strategy as AI becomes essential.
Microsoft Releases OptiMind for Turning Problems Into Math - Microsoft Research released OptiMind, an open model that converts plain-language optimization problems into mathematical formulas ready for solvers. It's designed for supply chain, scheduling, and logistics challenges.
Google Releases TranslateGemma for 55 Languages - Google released open translation models that outperform larger systems using fewer computing resources. The 12B model beats the 27B baseline and can run on consumer laptops or phones.
TRY THIS PROMPT (copy and paste into Claude,ChatGPT, or Gemini)
Difficult Conversation Simulator: Practice tough talks with AI roleplay, response options, and "what not to say" alerts
Build a fully functional Difficult Conversation Simulator as an interactive React app. Render the working application immediately - no code display.
**The conversation I'm dreading**: [What you need to discuss]
**Who it's with**: [Boss, partner, friend, family member, coworker, etc.]
Create these sections:
1. **Conversation Setup**
• Select conversation type:
💼 Asking for a raise/promotion
🚪 Quitting your job
💔 Ending a relationship
😤 Addressing someone's behavior
🙅 Setting a boundary
📉 Delivering bad news
🤝 Negotiating/disagreeing
👨👩👧 Family confrontation
• Describe the situation briefly
• What outcome do you want?
• What's making this hard?
• "Start Practice" button
2. **Roleplay Mode**
• Chat interface where AI plays the other person
• They respond realistically (not perfectly agreeable)
• Multiple response options for you:
- Option A: Direct approach
- Option B: Softer approach
- Option C: Question-based approach
• See how each choice plays out
• "They might say..." realistic reactions
• Reset and try different approaches
3. **Emotion Meter**
• Track emotional temperature during convo:
🥶 Cold/Defensive
😐 Neutral
🙂 Receptive
😤 Heated
💥 Explosive
• Visual gauge that shifts based on word choices
• "Tension rising" warnings
• Cool-down phrase suggestions
• "You just escalated things" alerts
4. **What NOT to Say**
• Red flag phrases that backfire:
🚨 "You always..." / "You never..."
🚨 "Calm down"
🚨 "That's not my problem"
🚨 "Whatever"
🚨 "I don't care"
🚨 "You're being ridiculous"
• Real-time detection if you type these
• Why it backfires explanation
• Better alternative suggested
• "Search Difficult Conversation Tips" button
5. **Power Phrases Library**
• Proven openers:
"I need to talk to you about something important"
"I've been thinking about..."
"Can we find a time to discuss..."
• De-escalation phrases:
"I hear what you're saying"
"Help me understand..."
"What would work for you?"
• Boundary statements:
"That doesn't work for me"
"I'm not comfortable with..."
• Copy button for each
6. **Scenario Branches**
• "What if they react with..."
- Anger → your response options
- Tears → your response options
- Denial → your response options
- Deflection → your response options
- Silence → your response options
• Practice each branch
• Best/worst case preparation
• "They surprised me with X" handler
7. **Conversation Script Builder**
• Your final game plan:
- Opening line
- Key points to make (max 3)
- Phrases to use
- Phrases to avoid
- Desired outcome statement
- Backup plan if it goes sideways
• One-page cheat sheet
• Print/save for reference
Design specs:
- Chat/messaging interface aesthetic
- Speech bubbles for roleplay
- Emotion meter as prominent gauge
- Branching path visualization
- Warm but serious palette (navy, soft gray, accent coral for warnings)
- Green highlights for good phrases
- Red alerts for danger phrases
- Smooth transitions between conversation branches
- "Choose your response" card selector
- Progress through conversation stages
When "Search Difficult Conversation Tips" is clicked, use web search to find communication psychology research and expert advice on navigating tough conversations.What this does: Lets you rehearse the conversation you've been avoiding. Practice different approaches, see how word choices affect the emotional temperature, get real-time warnings when you're about to say something that backfires, and build a script so you walk in prepared instead of winging it.
What this looks like:
WHERE WE STAND(based on today’s Quick Takes and Trending news)
✅ AI Can Now: Run continuously on factory floors doing hazardous electrical work that previously required human breaks and safety precautions.
❌ Still Can't: Succeed reliably on complex tasks - Anthropic's data shows AI success rates drop from 70% on simple work to below 50% on tasks requiring 5+ hours of human effort.
✅ AI Can Now: Turn plain English descriptions of business problems into mathematical optimization formulas ready for solving.
❌ Still Can't: Replace the human expertise needed to understand AI's work - higher-skill tasks show greater AI coverage but also require more human oversight to catch errors.
FROM THE WEB
RECOMMENDED LISTENING/READING/WATCHING
A documentary following Ray Kurzweil as he promotes his ideas about the singularity—the moment when AI surpasses human intelligence and transforms everything. The film includes interviews with supporters and critics, showing both sides of Kurzweil's radical optimism.
Kurzweil believes we'll merge with machines, defeat death, and expand throughout the universe. The documentary lets him make his case while also giving space to people who think he's delusional.
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.
