ChatGPT o3...just wow 🤯

Read time: 4 minutes

Hi everyone! it’s Sam—your guide to the AI frontier, back with another week of AI Simplified. Join me as I cut through the noise and break down the important news from this week in AI.

This was a packed week, and I’ve been speaking with so many people who are amazed at how quickly AI is being utilized in all areas of our daily routines. We’re living in a time where anything from a basic flip phone to cutting-edge Ray-Ban smart glasses can plug you into this powerful seemingly omniscient source of info. There is a lot to take in… But that’s why we’re here: to cut through the noise, highlight the gems, and give you a head start on the future. Buckle up, and let’s dive in!

🗞️ News:

🧠 OpenAI’s New o3 Model: AGI?

🥸 Google’s Gemini 2.0: “Flash Thinking Experimental”

👓️ Ray-Ban Meta Smart Glasses v11: AI in Your Eyes

🐣 Back to the Basics: How to use ChatGPT on WhatsApp

✍️ Prompt Tutorial: Plan an unforgettable party

⚡️ 3 AI tools to supercharge your productivity

📸 AI Images of the Day: A Subway… But Make It Surreal

🗞️ ^_News: 

🧠 OpenAI’s New o3 Model: AGI?

This week, OpenAI announced a preview of its brand-new “o3” models—an advance CEO Sam Altman called the “next phase” of AI.

So, what’s the big deal? These o3 models did really well on a test called the ARC-AGI benchmark, which measures how good an AI is at picking up new tasks (like a student quickly learning new skills). the question on everyone’s minds was , “had OpenAI had achieved “AGI,” or artificial general intelligence—the ultimate goal where an AI can do at least as well as humans in many different tasks.”?

The benchmark’s creator, François Chollet, says o3 is “very impressive” but still can’t solve some simpler tasks that a human would find easy. He does think this jump is a “big milestone” that could speed up serious research into AI’s capabilities.

Bottom line: o3 is a massive leap forward. AGI? Not yet.

What is amazing though is that o3 is now the new worst AI will ever be.

That alone should blow your mind.

As part of the unveiling, OpenAI also announced o3 mini, a smaller version of the model that’s been fine-tuned for specific jobs. For now, both versions are only available to safety researchers, who are testing them for any potential risks.

The release of o3 capped off OpenAI’s “12 Days of Shipmas,” a flurry of daily announcements. Here they are at a glance:

The 12 Days of Shipmas

1. Day 1: o1 & ChatGPT Pro

   • Introduction of the o1 model (an earlier AI version) and new premium features for ChatGPT.

2. Day 2: Reinforcement Fine-Tuning Research Program

   • OpenAI invited researchers and organizations to apply for a specialized program that helps advanced users tweak AI models for tough tasks.

3. Day 3: Sora

   • Debut of Sora, a video-generation tool that can create or remix content in different formats.

4. Day 4: Canvas

   • A shared workspace for writing, coding, and brainstorming alongside ChatGPT.

5. Day 5: ChatGPT in Apple Intelligence

   • New ways to integrate ChatGPT into Apple devices and apps.

6. Day 6: Advanced Voice with Video & Santa Mode

   • Showcased voice features for AI-driven videos and a fun holiday-themed “Santa Mode.”

7. Day 7: Projects in ChatGPT

   • Rolled out a feature to manage your AI-assisted tasks as “projects” within ChatGPT.

8. Day 8: Search

   • Announced improvements to AI-driven search capabilities, making it faster and more “intelligent.”

9. Day 9: Holiday Treats for Developers

   • A festive toolkit of new developer resources and goodies.

10. Day 10: 1-800-CHATGPT

    • Having a flip phone is not an excuse to not use AI anymore. Plus updates on how ChatGPT can work with external apps.

11. Day 11: Work with Apps

    • Showcased better integration between ChatGPT and third-party applications.

12. Day 12: o3 Preview & Safety Research Call

    • The grand finale, introducing the o3 models.

What’s Next?
While o3 isn’t publicly available yet, this really was a big moment because it shows that AI systems can continue to get more powerful even without scaling the LLM’s. They could soon tackle tougher questions, solve problems step by step, and find relevant information more intelligently. The crazy part? this is still the early days, and there’s a lot of work ahead.

You can learn more about it here →

🥸 Google’s Gemini 2.0: “Flash Thinking Experimental”

Google DeepMind made a splash with its new reasoning model: Gemini 2.0, labeled “Flash Thinking Experimental.” Picture this: an AI that literally hits pause, checks its work, and corrects itself before blurting out an answer.

• Fact-Checking Itself: Google says Gemini 2.0 has a built-in mechanism to cross-verify details. So instead of instantly spitting out an answer, it spends a few moments ensuring it’s not spewing nonsense. That might mean slower responses, but hey, quality over speed, right?

• Multi-Stage Reasoning: Early demos suggest it breaks big problems down into smaller steps, double-checks each one, and then compiles a final answer. Think of it like an AI that shows its work in math class, so if something goes wrong, you can see exactly which step messed up.

• Potential Impact: If Google nails this, it could mark a shift from AI that sounds confident-but-wrong to AI that’s more thoughtful, even if that means a slight delay. Personally, I’d take a 30-second wait for an accurate answer over a 5-second doozy.

Why I’m Excited:
Ever had ChatGPT or Bard say something that felt super convincing but turned out false? If you ever used it, your answer is definitely a yes. Gemini 2.0 aims to tackle that head-on. A game changer for reliable info from AI models.

You can read more about it here →

👓️ Ray-Ban Meta Smart Glasses v11: AI in Your Eyes

If you ever wanted a real-life HUD (heads-up display) from sci-fi movies, Ray-Ban and Meta are getting you closer.

• “Live AI” Feature: With firmware update v11, these glasses let you talk to Meta AI as you go about your day. It “remembers” past parts of the conversation, and can even see real-time video from your POV to provide context-sensitive info.
  Example: Imagine walking past a restaurant, and the glasses say, “Hey, you liked sushi last week. This place has a 4.5-star rating right now.”

• Live Translation: The update also adds instant translation among English, Spanish, French, and Italian. If you’re traveling, or just want to help a friend read a foreign menu, it’s like having a personal translator perched on your nose.

• Integration with Shazam: Spot a street musician playing a catchy tune, and Shazam can ID the song through your glasses in seconds.

My Take: This feels like a big leap in wearable tech. It’s not just a pair of fancy frames anymore—it’s like a real companion. If they can iron out privacy concerns (like people not wanting to be video-recorded unknowingly), it might be a serious use case for mainstream AR adoption. Also, the translation bit really creates frictionless travel—no more struggling with language barriers!

You can read more about it here →

 🐣 Back to the Basics: How to use ChatGPT on WhatsApp

You can now interact with ChatGPT via phone call or WhatsApp message. Here's how:

• Open WhatsApp, click on + at the top right, click on 'New Contact,' and add +1(800)2428478 as your contact.
  Alternatively, click on + and search for the +18002428478 number.

• Now open the chat in WhatsApp and type in any prompt.

• ChatGPT will answer within seconds.

• If you are in the US or Canada, you can even call ChatGPT on the same number.

 ✍️ Prompt Tutorial: Plan an unforgettable party

Objective: Plan an unforgettable birthday party (or any celebration), complete with theme ideas, food suggestions, entertainment options, and cost-saving strategies.

I want to host a small birthday party for around 25 people. I love the ‘Speakeasy’ vibe. Please suggest theme elements (decorations, music playlist, cocktail ideas), recommended party games, and approximate costs. My budget is around $300, and I’d like it to feel cozy but classy.

⚡️ 3 AI tools to supercharge your productivity

ElevenLabs

AI voices on demand, in multiple languages. Perfect for podcasters or content creators wanting to add quick voiceovers or dialogues to their videos without spending a fortune on voice actors.

Try it here →

Veo - Google DeepMind

Generate high-quality, cinematic videos with AI.

Try it here →

Wix AI Website Builder - GPT

AI-powered websites, instantly ready for business.

Try it here →

📸 AI Images of the Day: A Subway… But Make It Surreal

Imagine a Tokyo subway car at rush hour, but the floor is actually a shallow koi pond with vibrant orange fish. The overhead ads shift every few seconds between geometric neon art and calm, seaside landscapes. Soft, warm lighting envelops the space, giving it a cozy glow. Passengers stand on stepping stones, sipping matcha lattes and chatting as if this were totally normal.

🫡 We are here for you

Thanks for reading! When you signed up for this newsletter, our promise to you was to act as your bridge and help you understand what's happening in the world of AI.

What did you think of today's newsletter?

If you enjoyed it, please share it with a friend!

Reply to this email and let me know what you'd like to see more of.

Thanks for reading.

Until next time!

Sam Hadar ✍️