In partnership with
Effortless Tutorial Video Creation with Guidde
Transform your team’s static training materials into dynamic, engaging video guides with Guidde.
Here’s what you’ll love about Guidde:
1️⃣ Easy to Create: Turn PDFs or manuals into stunning video tutorials with a single click.
2️⃣ Easy to Update: Update video content in seconds to keep your training materials relevant.
3️⃣ Easy to Localize: Generate multilingual guides to ensure accessibility for global teams.
Empower your teammates with interactive learning.
And the best part? The browser extension is 100% free.
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
OpenAI's GPT-5 Improved a Biology Protocol by 79x
TLDR: GPT-5 created a novel molecular biology technique that improved cloning efficiency by 79x—the first time an AI model has produced verifiable new science in a physical laboratory.
The Story:
OpenAI partnered with Red Queen Bio, a biosecurity startup, to test whether GPT-5 could do more than ace benchmarks—whether it could actually improve real experiments. They gave the model a standard DNA cloning protocol and let it propose modifications, with human scientists running each experiment and feeding results back for iteration. Over multiple rounds, GPT-5 invented what it called "RAPF-HiFi": a new enzymatic method combining two proteins (RecA and gp32) that had never been paired this way before. The result was 79 times more sequence-verified clones from the same input DNA. Red Queen's chief scientist Nikolai Eroshenko said GPT-5 "went meaningfully beyond" existing research, showing "some glimpses of creativity."
Its Significance:
Biology has been AI's toughest domain because progress requires physical experiments, not just computation. Math proofs can be checked instantly; lab protocols need someone to actually run them. This work offers the first real evidence that frontier models can contribute genuinely novel ideas to experimental science—not just summarize literature or pass tests. OpenAI also built a robot that executes these AI-designed protocols autonomously, hinting at a future where AI proposes experiments and machines run them continuously. That said, the researchers were careful: "It's not a foundational breakthrough in molecular biology," said OpenAI's Miles Wang. "But it's accurate to call it a novel improvement, because it hasn't been done before."
QUICK TAKES
The story: OpenAI released GPT Image 1.5, a new image generation model that's up to four times faster than the previous version. The biggest improvement is editing—when you ask it to change something in a photo, it now keeps the rest of the image intact instead of regenerating everything. It also handles text in images much better, which has been a weakness for AI image tools. The release is part of OpenAI's "code red" push to catch up after Google's Nano Banana Pro image generator went viral last month.
Your takeaway: Speed and editing consistency have been major frustrations with AI image tools. If OpenAI delivers on these promises, it could pull users back from Google's Gemini, which had been gaining ground with 650 million monthly users.
The story: Meta's Ray-Ban smart glasses received a software update adding two new AI features. "Conversation Focus" uses AI to filter out background noise and amplify the voice of whoever you're talking to. A new Spotify feature lets you say "play a song that matches this view" and the glasses will analyze what you're looking at and pick music to fit the scene.
Your takeaway: These features show how AI is moving beyond screens and into wearables. The noise-filtering could genuinely help in loud environments, though having AI choose your music based on what it "sees" raises questions about how much context we want our devices to have.
The story: The Washington Post rolled out AI-generated podcasts to readers even though internal testing showed major problems. Tests found between 68% and 84% of AI-written scripts were unpublishable due to errors, made-up quotes, and wrong information. Staff complained loudly, but leadership pushed forward anyway, calling it a "Beta" product and normal testing.
Your takeaway: This shows companies are rushing AI products to market even when their own teams say they're not ready. The "Beta" label is becoming a way to release broken tools while shifting blame to the technology.
The story: Chinese tech company Kuaishou released Kling 2.6, the first major AI video tool that creates visuals and sound at the same time. Instead of making silent video and adding audio later, it generates dialogue, sound effects, and background noise in one step. It works in both Chinese and English and creates 10-second clips at 1080p quality.
Your takeaway: This removes a major headache in AI video creation. Syncing audio to video has been tedious manual work—now it happens automatically, which could speed up content creation significantly.
The story: The UK government is pushing Apple and Google to build nudity-detecting AI directly into phone operating systems. Users would need to verify their age using ID or biometrics to turn off the filter. The government says it's "encouraged" rather than required—for now. This comes as part of the UK's Online Safety Act.
Your takeaway: This would put AI scanning on every phone by default, raising extreme privacy concerns similar to Apple's abandoned photo-scanning plans from 2021. What starts as "encouraged" often becomes required and the UK has been called out for an increasing number of long prison terms based on speech.
The story: After Mozilla's new CEO announced Firefox would become an "AI-first" browser, the team behind Waterfox (a Firefox alternative) declared they won't add large language models to their browser. Waterfox's founder argued that browsers should "serve you, not think for you" and that AI creates a "black box" users can't audit or trust.
Your takeaway: As major browsers race to add AI features, some alternatives are betting users will want simpler, more transparent tools. This split could define browser competition for years.
TOOLS ON OUR RADAR
💰 Wave
Freemium: Send unlimited professional invoices, track expenses, and manage your bookkeeping—all for free with this small business accounting tool trusted by over 2 million users.✍️ Hemingway Editor
Freemium: Paste your writing and instantly see which sentences are hard to read, where you're using passive voice, and get a readability grade—no signup required.🔖 Raindrop.io
Freemium: Save unlimited bookmarks across all your devices, organize them into visual collections, and find anything instantly with powerful search and tagging.🖥️ ScreenRec
Freemium: Record your screen with webcam and audio for free with no time limits and no watermarks, then share instantly via a private cloud link.
TRENDING
Microsoft Says Windows 11 AI Will Ask Before Accessing Your Files — After user complaints, Microsoft confirmed that AI tools in Windows 11 will ask permission before reading personal folders like Documents and Downloads.
500+ Creators Form Coalition to Push for AI Rules — Actors, writers, and directors including Natalie Portman and Guillermo del Toro launched a group demanding standards for how AI companies use creative work, with plans for public pressure and possible legal action.
Google Launches Experimental AI Assistant for Email and Calendar — Google's new "CC" agent connects to Gmail, Calendar, and Drive to send you daily briefings and draft emails based on your schedule.
Looki Launches $212 AI Wearable Camera in China — The 32-gram Looki L1 records your day and uses AI to automatically create highlight videos without any editing.
OpenAI in Talks for $10B+ From Amazon — OpenAI is discussing a deal that would bring at least $10 billion from Amazon and have OpenAI use Amazon's Trainium AI chips, potentially pushing its valuation past $500 billion.
TRY THIS PROMPT (copy and paste into Claude or Gemini)
Feynman Technique Trainer: Test if you truly understand something by explaining it in simple terms—gaps reveal what you still need to learn
Build me an interactive Feynman Technique Trainer as a React artifact that helps you learn deeply by forcing simple explanations and identifying knowledge gaps.
The console should include these sections:
1. **Topic Setup** - What are you learning?
• Concept or topic input (e.g., "photosynthesis," "blockchain," "supply and demand")
• Subject area: Science, Technology, Business, Math, History, Philosophy, Other
• Your current understanding: Beginner, Intermediate, Advanced
• Learning goal: Quick overview, Deep mastery, Teach others, Exam prep
• "Start Learning" button
2. **Step 1: Study & Take Notes** - Initial learning:
• Large text area for your initial notes
• "Search for Resources" button to find explanations and videos
• Key terms highlighter
• Timer showing study duration
• "I'm ready to explain" button when done
3. **Step 2: Explain Simply** - The core challenge:
• **Notebook interface** (lined paper aesthetic)
• Prompt: "Explain [topic] as if teaching a 12-year-old"
• Large writing area for your explanation
• Real-time feedback sidebar showing:
- Word count
- Complexity score (reading level)
- Jargon detector (highlights technical terms)
- Analogy counter (encourages metaphors)
• "Check My Explanation" button
4. **Step 3: Gap Identifier** - Find what you don't know:
• AI analyzes your explanation and identifies:
- **Missing concepts** (important parts you skipped)
- **Vague language** ("it works by doing stuff")
- **Circular definitions** (using the word to define itself)
- **Jargon dependency** (terms you didn't simplify)
- **Logical jumps** (skipped steps in reasoning)
• Each gap marked with:
- What's missing
- Why it matters
- What to study next
• Gap severity: Minor, Moderate, Critical
• Visual progress bar (understanding completeness %)
5. **Step 4: Review & Simplify** - Fill the gaps:
• For each identified gap:
- "Study This Gap" button (searches for targeted info)
- Note-taking area specific to this gap
- Checkbox when you've learned it
• Re-explain prompt with focus on gaps
• Side-by-side: Your first explanation vs. revised explanation
• Track improvement (fewer gaps, lower complexity)
6. **Analogy Workshop** - Make it relatable:
• "Generate Analogy Ideas" for your topic
• Analogy builder:
- [Concept] is like [everyday thing]
- Both have [common property]
- The difference is [key distinction]
• Test analogies: Are they accurate? Misleading?
• Save best analogies to use in explanations
7. **Complexity Meter** - Simplicity score:
• Flesch-Kincaid reading level (target: 6th-8th grade)
• Sentence length average
• Technical term density
• Passive voice detector
• "Simplify This" suggestions for complex sentences
• Before/after comparison when simplified
8. **Teaching Simulator** - Practice explaining:
• "Virtual Student" asks questions about your topic:
- Basic questions (testing fundamentals)
- "Why?" questions (testing deeper understanding)
- "What if?" scenarios (testing application)
• You answer in the chat
• Feedback on answer quality
• If you can't answer → that's a gap to fill
9. **Mastery Dashboard** - Track progress:
• Understanding score (0-100%)
• Gaps filled vs. remaining
• Complexity improvement over time
• Number of revision cycles
• Time spent learning
• "Ready to teach?" assessment
• Export your final simple explanation
Make it look like a learning notebook with:
• Notebook/journal aesthetic (lined paper, handwritten feel for headings)
• Warm, educational color palette (cream, soft blue, pencil gray)
• Clear visual hierarchy for steps
• Progress indicators throughout
• Encouraging feedback ("You're getting it!")
• Gap highlights in yellow (like a highlighter)
• Green checkmarks when gaps are filled
• Clean, readable typography
• Minimal distractions for focus
When I click "Search for Resources" or "Study This Gap," use web search to find simple explanations, educational videos, analogies, and teaching resources for the specific concept or knowledge gap.What this does: Uses the Feynman Technique (explain simply to learn deeply) to expose knowledge gaps—forcing you to identify exactly what you don't understand so you can study strategically instead of just re-reading everything.
What this looks like:
WHERE WE STAND
✅ AI Can Now: Generate video with synchronized dialogue, sound effects, and background audio in a single step—no manual audio editing required.
❌ Still Can't: Reliably write accurate news content, with one major outlet finding up to 84% of AI-generated scripts had errors or made-up information.
✅ AI Can Now: Filter out background noise in real-time to help you hear conversations better through smart glasses.
❌ Still Can't: Be fully audited or explained—even browser makers say current AI works like a "black box" that users can't inspect or verify.
✅ AI Can Now: Connect your email, calendar, and files to give you personalized daily briefings and draft messages automatically.
❌ Still Can't: Access your personal files without raising privacy alarms—companies are adding permission systems after users pushed back.
FROM THE WEB
RECOMMENDED LISTENING/READING/WATCHING
SHORT FILM: More (1998)
In a gray world, people plug into machines to experience "More". One inventor creates his own device, but what starts as wonder gradually becomes something else.
This short is only six minutes. Director Mark Osborne(who went on to co-direct Kung Fu Panda) doesn't explain much, and doesn’t need to. This one is about the feeling that you get at the very end.
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.
