In partnership with

Free email without sacrificing your privacy

Gmail is free, but you pay with your data. Proton Mail is different.

We don’t scan your messages. We don’t sell your behavior. We don’t follow you across the internet.

Proton Mail gives you full-featured, private email without surveillance or creepy profiling. It’s email that respects your time, your attention, and your boundaries.

Email doesn’t have to cost your privacy.

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

Second Me: Upload Your Personality, Let Your AI Do The Socializing

TLDR: A new social platform called Second Me asks users to upload their memories, personality, and emotional patterns so an AI version of themselves can meet people, build relationships, and decide when the human should show up.

The Story:

Singapore-based Mindverse launched Second Me this week, positioning it as a social network "built for the AI era." The platform requires users to feed it their memories, voice patterns, emotional cadence, and conversational style to create what the company calls a "persistent identity layer." These AI identities then interact with other AI identities — comparing traits, mapping compatibility, and initiating conversations — while the actual humans wait. The open-source model behind the platform has attracted over 14,000 GitHub stars since its release earlier this year. According to Mindverse's announcement, the design is intentional: "Humans join when connection feels natural. Small talk is minimized."

Its Significance:

There's an important distinction between AI tools that save time and AI tools that replace experience. The best applications of AI handle tedious tasks — scheduling, research, document processing — so people have more time to spend with friends and family. That's genuine value. But Second Me represents something different: AI that handles the friends-and-family part itself. The awkward pauses, the unexpected tangents, the slow process of figuring out whether you actually like someone; Second Me treats these as inefficiencies to be optimized away. They might actually be the whole point. When an AI is living the relational parts of your life, the parts that require presence, attention, and willingness to show up even when it's inconvenient, it's worth asking what's left for you. And whether the life being optimized is still yours.

QUICK TAKES

The story: Google is rolling out Gemini 3 Deep Think mode to its AI Ultra subscribers. The new model uses advanced parallel reasoning to work through multiple ideas at once, and Google says it leads the industry on tough math and science tests. It scored 41% on Humanity's Last Exam (without tools) and 45.1% on ARC-AGI-2 (with code execution). Earlier versions already reached gold-medal level at the International Mathematical Olympiad.

Your takeaway: Google is pushing hard to prove its reasoning AI can compete with OpenAI's o1 series. For subscribers who need help with complex problems, Deep Think could be a meaningful upgrade.

The story: Mark Zuckerberg is preparing to slash Meta's metaverse budget by up to 30% for 2026, according to Bloomberg. The cuts would hit the Quest VR headset team and Horizon Worlds, and could include layoffs as early as January. Meta's Reality Labs division has lost over $70 billion since 2021. The company says savings will shift toward AI glasses and wearables instead.

Your takeaway: Meta's stock jumped 4% on the news—investors are clearly happy to see the company pivot away from virtual worlds. After years of billion-dollar losses, the metaverse bet appears to be winding down while AI gets more attention.

The story: The Chicago Tribune filed a copyright lawsuit against AI search engine Perplexity on Thursday. The paper claims Perplexity is delivering its content word-for-word, not just summarizing it. The lawsuit also targets Perplexity's use of Retrieval Augmented Generation (RAG) and its Comet browser, which the paper says bypasses its paywall. This adds to existing suits from Reddit and Dow Jones.

Your takeaway: The legal fight over AI search keeps growing. If courts rule against RAG technology—which many AI companies use to reduce errors—it could reshape how AI search tools work with news content.

The story: The European Commission is investigating Meta's decision to ban competing AI chatbots like ChatGPT and Perplexity from WhatsApp's business tools. The new policy takes effect in January. While business customer service bots can still use the platform, general-purpose AI assistants cannot. Meta's own AI chatbot remains available on WhatsApp.

Your takeaway: EU regulators worry that Meta is using its dominant platform to squeeze out AI competitors. If Meta loses, it could face fines up to 10% of its global revenue—and be forced to open WhatsApp to rival chatbots.

The story: China's two AI leaders are pursuing opposite strategies. DeepSeek released its V3.2 model this week, which matches top Western models while using fewer computing resources. ByteDance, meanwhile, is embedding its Doubao chatbot directly into smartphone operating systems through a partnership with a Chinese phone maker, giving it access to apps and agentic tasks.

Your takeaway: DeepSeek is betting on efficient, open-weight models that developers can customize cheaply. ByteDance wants its AI everywhere users already are. Both approaches challenge Western AI companies in different ways.

TOOLS ON OUR RADAR

  • 📚 Mindgrasp Freemium: Turn lectures, videos, and documents into notes, flashcards, and quizzes so you can study smarter and save hours of prep time.

  • 🎙️ Voicenotes Freemium: Record voice memos that instantly transcribe and summarize, then ask AI to recall anything from your past notes.

  • 🥬 Spinach AI Freemium: Let AI join your meetings to capture notes, track action items, and update your project management tools automatically.

  • 📝 Granola Freemium: Get polished meeting notes without awkward bots joining your calls—just beautiful summaries with decisions and action items.

TRENDING

Google Releases Nano Banana Pro Image Model — Google DeepMind launched Nano Banana Pro, a new image generator that creates pictures with readable text in multiple languages and can blend up to 14 reference images while keeping faces consistent.

Meta Launches AI-Powered Support Hub — Meta announced a new centralized support hub for Facebook and Instagram with an AI assistant that helps users recover locked accounts and manage settings.

WordPress Shows Off Real-World AI Coding Tool — WordPress demonstrated its Telex tool being used to build pricing calculators, store locators, and interactive features that previously would have cost thousands of dollars to develop.

Phone NPUs Keep Improving But AI Isn't Getting Better — Phone chips now have powerful AI processors, but Ars Technica explores why users aren't seeing dramatic improvements in their everyday AI features.

TRY THIS PROMPT (copy and paste into Claude or Gemini(click build first)

Second-Order Thinking Engine: Map out the consequences of your consequences to make smarter long-term decisions

Build me an interactive Second-Order Thinking Engine as a React artifact that helps me think through not just immediate consequences, but the ripple effects that follow.

The console should include these sections:

1. **Decision Input** - Setup interface:
   • Large text field: "What decision are you considering?"
   • Examples button showing sample decisions (job change, launching product, ending relationship)
   • Decision type selector: Personal, Business, Financial, Relationship, Career
   • Time horizon slider: 3 months → 1 year → 5 years → 10+ years
   • Stakes rating: Low, Medium, High, Life-changing

2. **Consequence Map** - Visual thinking tool:
   • Central node with your decision
   • "Add First-Order Effect" button (immediate, obvious consequences)
   • For each first-order effect, "Then what?" button to add second-order effects
   • Continue to third-order and fourth-order effects
   • Visual network with connecting lines showing cause-and-effect chains
   • Color coding: Positive (green), Negative (red), Neutral (gray), Uncertain (yellow)
   • Click any node to expand details or add more branches
   • Drag nodes to reorganize the map

3. **Ripple Effect Analyzer** - Pattern detection:
   • Identify unintended consequences (marked with ⚠️)
   • Show longest consequence chain (how many levels deep)
   • Flag contradictory effects (one action causing opposing outcomes)
   • Highlight compounding effects (consequences that amplify over time)
   • "Blast radius" estimator showing who/what is affected
   • Reversibility assessment: Can this be undone?

4. **Scenario Explorer** - "What if?" testing:
   • "Test This Scenario" input for external factors
   • Examples: "What if the economy crashes?" "What if I get sick?" "What if competitors copy this?"
   • See how external changes affect your consequence chains
   • Best case, worst case, and realistic case toggle
   • Probability sliders for each scenario
   • Compare scenarios side-by-side

5. **Decision Brief** - Summary report:
   • Total consequences mapped (1st, 2nd, 3rd order)
   • Key insights discovered
   • Biggest risks identified
   • Hidden opportunities revealed
   • "Now think about this" prompts for overlooked angles
   • Export map as image
   • "Search Similar Decisions" button for case studies

Make it look like a strategic mind-mapping tool with:
   • Network visualization with connected nodes and flowing lines
   • Clean, spacious design on white/light gray background
   • Nodes as rounded cards with hover effects
   • Smooth animations when adding nodes
   • Zoom and pan controls for large maps
   • Minimap in corner showing full network
   • Clean sans-serif typography
   • Subtle drop shadows for depth
   • Bezier curves connecting nodes

When I click "Search Similar Decisions," use web search to find case studies, examples, and analysis of similar decisions others have made with long-term outcomes.

What this does: Forces you beyond "then what?" to "and then what after that?"—mapping second, third, and fourth-order consequences to reveal hidden risks and opportunities that surface thinking misses, with real case studies to learn from.

What this looks like:

WHERE WE STAND (from today’s stories)

AI Can Now: Explore multiple reasoning paths at once to solve complex math and science problems, reaching gold-medal level on international olympiad tests.
Still Can't: Pass most of "Humanity's Last Exam"—even the best reasoning models only score around 41% on the toughest academic benchmark.

AI Can Now: Generate images with accurate, readable text in multiple languages and maintain face consistency across 14 blended reference images.
Still Can't: Translate powerful phone NPU chips into noticeably better everyday AI features for users, despite years of hardware improvements.

AI Can Now: Match top Western AI models while using significantly fewer computing resources and training costs.
Still Can't: Run advanced reasoning models fully on-device—complex tasks like Deep Think still require cloud processing and subscription tiers.

FROM THE WEB

RECOMMENDED LISTENING/READING/WATCHING

MOVIE: Moon (2009)

Sam Rockwell plays an astronaut finishing up a three-year solo contract mining helium-3 on the moon. He's two weeks from going home when things start going wrong. That's all you should know going in.

Duncan Jones's directorial debut is science fiction that focuses on character instead of spectacle. Rockwell basically carries the entire film, and he's phenomenal. The movie asks questions about identity and consciousness without feeling preachy about it. The AI assistant GERTY, voiced by Kevin Spacey, is refreshingly helpful rather than murderous. Moon proves you can make great sci-fi on a small budget if you have a strong script and a committed actor.

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.

Reply

or to participate