In partnership with
This newsletter you couldn’t wait to open? It runs on beehiiv — the absolute best platform for email newsletters.
Our editor makes your content look like Picasso in the inbox. Your website? Beautiful and ready to capture subscribers on day one.
And when it’s time to monetize, you don’t need to duct-tape a dozen tools together. Paid subscriptions, referrals, and a (super easy-to-use) global ad network — it’s all built in.
beehiiv isn’t just the best choice. It’s the only choice that makes sense.
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
TLDR: OpenAI launched a browsable App Directory inside ChatGPT and opened submissions for developers who want their apps reviewed and published—marking the company's second attempt at building an app ecosystem after the GPT Store failed to gain traction. This second attempt is a move towards something closer to a new type of operating system.
The Story:
The directory went live December 17 across web, iOS, and Android for all logged-in users outside the EU, UK, and Switzerland. Launch partners include Spotify, Zillow, Canva, Coursera, Booking.com, and Dropbox, with Uber, Target, and Peloton coming later. Apps work differently than the old custom GPTs—they render actual UI elements inline (maps, playlists, videos) and connect to backends for real functionality. Users can find them through the tools menu, @-mention them in chats, or let ChatGPT suggest them based on conversation context. Apps is built on the Model Context Protocol, an open standard that lets AI systems connect to external tools and data. Anthropic created it in late 2024, and it's now jointly managed by Anthropic, OpenAI, and Block under the Linux Foundation.
Its Significance:
This is OpenAI's do-over. The GPT Store launched in January 2024 with over 3 million custom GPTs, but they were essentially bundled prompts—as base models improved, users had no reason to hunt for specialized chatbots. The new approach adds real interactivity and commerce potential through the Agentic Commerce Protocol, an open standard co-developed with Stripe that enables in-chat checkout (already live with Etsy sellers, with Shopify merchants coming). First approved apps roll out early 2026. If this works, ChatGPT becomes less of a chatbot and more of an operating system; a single interface where you book travel, order groceries, manage files, and buy products without ever opening another app.
QUICK TAKES
The story: Google released Gemini 3 Flash, a faster and cheaper version of its Gemini 3 Pro model. The new model is now the default in the Gemini app and Google Search's AI Mode worldwide. It scored 33.7% on the Humanity's Last Exam benchmark and 78% on SWE-bench coding tests—matching or beating much larger models. Pricing is $0.50 per million input tokens and $3.00 per million output tokens.
Your takeaway: Google is making its most powerful AI accessible to everyone for free through its apps, while keeping the model cheap enough for developers to build with at scale.
The story: Zoom announced its AI system scored 48.1% on Humanity's Last Exam, beating Google's previous record of 45.8%. But critics quickly pointed out that Zoom didn't build its own AI model—it combined API calls to Gemini, GPT, and Claude to get the result. Some called it a smart engineering approach, while others accused Zoom of taking credit for other companies' work.
Your takeaway: This sparks an important debate about what counts as AI achievement. As companies combine multiple models to get better results, the industry may need new ways to measure and credit AI progress.
The story: CodeRabbit analyzed hundreds of open-source code submissions and found AI-generated code contained 1.7 times more issues than human-written code. The problems included 75% more logic errors, 3x more readability issues, and up to 2.74x more security vulnerabilities. Error handling gaps were nearly twice as common in AI contributions.
Your takeaway: AI coding tools are getting popular fast, but this data shows they still need human oversight. Teams using AI assistants should add extra review steps to catch the mistakes these tools tend to make.
TOOLS ON OUR RADAR
⏱️ Clockify
Free: Track billable hours across unlimited projects and team members with one-click timers, timesheets, and detailed reports—completely free forever.🎨 Photopea
Free: Edit photos like a pro with this browser-based Photoshop alternative that opens PSD files, supports layers and masks, and requires no download or signup.🌐 Carrd
Freemium: Build beautiful one-page websites for your portfolio, landing page, or link-in-bio in minutes—free for up to 3 sites with no coding needed.✂️ Remove.bg
Freemium: Instantly erase backgrounds from photos using AI—just upload an image and download a clean cutout in seconds, perfect for product shots and headshots.
TRENDING
Luma Launches Video Model That Fills in the Middle – Luma's new Ray3 Modify model lets creators provide a start and end frame, then generates the video footage in between while preserving actor movements and timing.
Amazon Adds AI Greeter to Ring Doorbells – A new Alexa+ feature called Greetings can identify visitors by their appearance and respond accordingly—directing delivery drivers, turning away salespeople, or taking messages from friends.
YouTube Lets Creators Build Games with AI – YouTube Gaming launched Playables Builder, a tool that uses Gemini 3 to let creators make simple games without coding knowledge.
Startup Uses Music Production Background to Build Spatial AI – A Romanian founder is applying his experience in data engineering and music production to develop new approaches to spatial intelligence and computer vision.
TRY THIS PROMPT (copy and paste into Claude or Gemini)
Value Hierarchy Explorer: Discover your true priorities through an interactive game that reveals what you actually value most
Build me an interactive Value Hierarchy Explorer as a React artifact that helps you identify and rank your core values through choices and trade-offs.
The console should include these sections:
1. **Value Discovery** - Find your values:
• Browse value library (40+ common values):
- Achievement, Adventure, Authenticity, Balance, Belonging
- Creativity, Family, Freedom, Growth, Health
- Honesty, Independence, Justice, Knowledge, Leadership
- Love, Loyalty, Peace, Recognition, Security
- Service, Spirituality, Stability, Success, Tradition
- Wealth, Wisdom, and more
• "What matters to you?" question prompts
• Select 10-15 values that resonate
• Add custom values if needed
• Quick-pick by life stage (student, parent, entrepreneur, retiree)
2. **The Sorting Game** - Rank through choices:
• **Round 1: Binary Choices**
- Present two values at a time
- "Which matters more?" forced choice
- Example: "Freedom vs. Security"
- Click to choose, repeats with different pairs
- Automatically ranks based on choices
• **Round 2: Hard Trade-offs**
- Scenario-based questions:
- "You can have a high-paying job you dislike OR fulfilling work with less money"
- "Time with family OR career advancement opportunity"
- Your choices reveal true priorities
• **Round 3: Sacrifice Test**
- "If you had to give up one, which would it be?"
- Forces you to eliminate lower priorities
- Repeat until top 5 values remain
3. **Value Pyramid** - Visual hierarchy:
• Pyramid or stacked layers showing your ranked values:
- Top tier (1-2): Core non-negotiables
- Mid tier (3-5): Important but flexible
- Base tier (6-10): Nice to have
• Drag values between tiers to adjust
• Color-coded by category (relationships, achievement, security, etc.)
• Click any value to see your definition of it
• "Lock in" top 3 values
4. **Conflict Detector** - Find tensions:
• Identify value pairs that conflict:
- "Adventure vs. Security"
- "Independence vs. Belonging"
- "Achievement vs. Balance"
• Real-life scenarios where they clash:
- Career decisions
- Relationship choices
- Lifestyle trade-offs
• "How do you typically resolve this?" reflection prompts
• Strategies for balancing competing values
5. **Decision Filter** - Apply your values:
• Test a current decision against your values:
- Input the decision you're facing
- Score how each option aligns with top 5 values
- Visual alignment chart
- "Value fit score" for each option
• Example questions:
- "Does this job match my values?"
- "Is this relationship aligned with what I value?"
- "Does this purchase reflect my priorities?"
• Clear recommendation based on values
6. **Life Audit** - Check alignment:
• Rate current life areas (1-10):
- Career
- Relationships
- Health
- Finances
- Personal growth
- Leisure/fun
• Compare against your top values
• Gap analysis: "You value X but spend little time/energy on it"
• Misalignment warnings (high investment, low value match)
• Rebalancing suggestions
7. **Value Statement Generator** - Define yourself:
• Create your personal value statement:
- "I prioritize [top value] because..."
- "I'm willing to sacrifice [lower value] for [higher value]"
- "When making decisions, I ask myself..."
• Generated statement based on your hierarchy
• Editable and exportable
• "Share with partner/team" option for alignment discussions
8. **Revisit & Evolve** - Track changes:
• Save multiple value hierarchies over time
• Compare "You 6 months ago" vs. "You now"
• See how major life events shifted values
• Quarterly review reminders
• "Has anything changed?" check-in questions
Make it look like a thoughtful self-discovery tool with:
• Pyramid/hierarchy visualization as centerpiece
• Soft, contemplative color palette (muted purples, blues, earth tones)
• Draggable value cards with smooth animations
• Clean, spacious layout (breathing room for reflection)
• Progress through sorting rounds
• Encouraging, non-judgmental language
• Personal journal aesthetic
• Gentle transitions between sections
• "Pause and reflect" moments built in
When I click "Search Value Resources," use web search to find values clarification exercises, philosophical frameworks on values, and personal development resources for living according to your values.What this does: Reveals your actual value hierarchy (not what you think you should value) through forced choices and trade-off scenarios—giving you a clear decision-making filter and exposing when your life doesn't match what you truly care about.
What this looks like:
WHERE WE STAND (from today’s stories)
✅ AI Can Now: Combine multiple AI models together to score higher on difficult tests than any single model could alone.
❌ Still Can't: Build and train its own models from scratch—it still relies on combining work that others have already created.
✅ AI Can Now: Generate video footage that smoothly connects two images while keeping a person's movements and expressions consistent.
❌ Still Can't: Reliably tell the difference between a delivery driver and a friend who happens to be wearing a work uniform.
✅ AI Can Now: Write working code faster than humans and help developers build software more quickly.
❌ Still Can't: Match human programmers on code quality—AI-generated code has nearly twice as many errors and security problems.
RECOMMENDED LISTENING/READING/WATCHING
MOVIE: 2001: A Space Odyssey (1968)
A mission to Jupiter goes wrong when HAL 9000, the ship's AI, decides the crew is jeopardizing the mission. But that's just one part of Kubrick's meditation on evolution, intelligence, and humanity's place in the universe. The film demands patience, but it rewards attention with images and ideas that have influenced every space film since.
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.
