Browse previous editions of the newsletter by clicking here

Hey AI team! Here we go with a new edition of Five Minute AI ⏰

On today’s menu:

• OpenAI unveils Sora videos made by professionals, showing striking realism.

• OpenAI’s o1 model earns an estimated score of 120 on Mensa Norway’s IQ test.

• Sam Altman talks about the rapid evolution of AI, from the o1 model to future autonomous agents.

• Huawei introduces a triple-screen foldable smartphone, pushing the limits of mobile design.

Ready? Let’s go...

DEFINITION OF THE DAY

What is a Prompt

In AI, a prompt is a piece of text written by the user to guide an artificial intelligence model through a specific task. That text can take the form of questions, instructions or descriptions, and it lets the AI understand what’s expected of it, whether that’s generating an answer, creating content or solving a problem. The clearer and more detailed the prompt, the more relevant and accurate the AI’s output will be.

THE MAIN STORY

📱 A revolutionary triple-screen smartphone

Huawei just unveiled a foldable phone with a double hinge and a triple screen, a significant step forward for flexible mobile devices. This innovative model stands out not only for its unique design but also for its premium positioning.

Details:

• Starting price: 19,999 yuan (about $2,809) for the 256 GB model

• OLED screen measuring 6.4 inches folded, expanding to 10.2 inches fully unfolded

• Possible intermediate 7.9-inch configuration

• 16 GB of RAM across all models

• 5,600 mAh battery with 66W wired fast charging

Why I’m telling you about this
This smartphone is an interesting evolution in the world of foldables. It offers unprecedented flexibility, letting users switch between screen sizes depending on what they need. Its high price may limit it to a niche audience, but it shows the growing possibilities in design and functionality across the mobile industry. With 3.7 million preorders in China already, this model is generating real interest, even if its availability outside China remains uncertain.

🧠 AI’s evolution according to Sam Altman: from o1 to agents

OpenAI CEO Sam Altman recently shared his thoughts on the rapid evolution of artificial intelligence at T-Mobile’s Capital Markets Day. His remarks highlight the advanced capabilities of the new o1 model and hint at a dramatic acceleration in the development of AI agents.

Details:

• o1 is presented as the first AI system with advanced reasoning capabilities, outperforming previous models at solving complex problems.

• Altman compares o1’s current state to the “GPT-2 stage” of reasoning models, suggesting this is just the beginning and that the GPT-4 equivalent is on the horizon.

• He outlines five levels of AI development, placing o1 at level 2 (reasoners) and hinting that level 3 (agents) could follow “relatively quickly.”

Why I’m telling you about this:
Altman’s comments suggest that the time between o1 and AI systems fully capable of acting autonomously will be considerably shorter than the jump from GPT-4 to o1. That fast pace could have major implications for many industries, paving the way for even more sophisticated and autonomous AI applications in the near future. It’s crucial to follow these developments to anticipate the potential impact on our lives and our industries.

For those who want to dig deeper, Altman’s full talk is available on YouTube (from 51:00 to 1:06:00).

Help Five Minute AI get the word out! 🚀

Share our passion and earn exclusive rewards.

By clicking the Share button below, you can invite your friends to join our community and unlock unique gifts with every new signup!

1️⃣ The Ultimate ChatGPT Manual, with tips only insiders know. This guide will help you boost your productivity and simplify your daily tasks → 🎁 Unlock with 3 friends signed up

2️⃣ The “Five Minute AI” collector mug: Featuring our adorable mascot Pixl enjoying its daily read → 🎁 Unlock with 15 friends signed up

Ready to spread AI knowledge? Let’s go! 🌟

LEARN TO MASTER AI

🧠 Optimize your prompts for OpenAI’s new o1 model

OpenAI’s o1 model offers new reasoning capabilities that call for a different approach to writing prompts. Here are a few tips to get the most out of this innovation 😀 

Example of using o1 to write a complex poem

1. Access o1: Log in to ChatGPT (paid subscription required) and select o1-preview or o1-mini from the model dropdown menu.

2. Simplify your prompts: Go for simple, direct instructions. Avoid the “Chain-of-Thought” style instructions you may have used with GPT-3 or GPT-4.

3. Structure with XML tags: Use XML tags to clarify your requests. Don’t hesitate to test the model with complex or philosophical questions to explore its capabilities.

4. Analyze the thought process: After each answer, click the “thinking” indicator to understand the model’s line of reasoning.

To deepen your mastery of o1, experiment with different types of requests and check out OpenAI’s new guide “Advice on prompting”. These techniques will let you take full advantage of o1’s advanced reasoning capabilities, opening up new possibilities for your AI projects.

WHAT THE RESEARCH SAYS

🧠 OpenAI’s o1 model: an IQ above the human average?

OpenAI’s new artificial intelligence model, called “o1,” appears to have reached a significant milestone in AI. According to recent tests, the model reportedly achieved impressive results on Mensa Norway’s IQ test, suggesting cognitive abilities potentially above the human average.

Details:

• Estimated score of 120 on Mensa Norway’s IQ test

• 25 correct answers out of 35 questions

• Strong performance on complex visual and logic puzzles

• Similar results on never-before-seen questions, ruling out simple memorization

Why I’m telling you about this
These results, though not officially confirmed by OpenAI, could mark an important step toward artificial general intelligence (AGI). If this performance holds up, it would indicate that the o1 model has reached stage 2 on OpenAI’s scale of progress toward AGI. This crucial step suggests a greater ability for AI to reason and solve complex problems, opening the door to new potential applications across many fields.

THE LATEST TRENDING TOOLS

🖼️ Reshot offers a new way to liven up your portraits. This AI-powered photo editing tool lets you subtly change the facial expression and head position in your shots, whether they’re real or AI-generated.

One of Reshot’s strengths is its ability to optimize YouTube thumbnails. By amplifying emotions and adjusting the angle of the face, you can potentially boost the click-through rate on your videos. The tool aims to create more eye-catching visuals without completely distorting the original image.

Reshot costs $19 per month, making it an affordable option for content creators who want to improve their visual presence online simply and effectively.

💡 Other tools:

• 🎬 Kling 1.5 brings notable improvements to video generation with better dynamic quality and more accurate prompt adherence. With the motion brush, you can choose the movement you want during generation. Yet another huge new development in video generation has just arrived!

• ✂️ Choppity lets you clip the key moments in videos based on visual, audio and emotional cues.

• 📚 Trupeer makes it easy to create guides and videos for any type of product, simplifying technical documentation.

• 🧞 CX Genie offers no-code AI solutions to improve customer support efficiently.

• 📝 Essay Grader uses AI to grade essays and offer suggestions for improvement. English-only for now but very interesting, maybe an idea worth developing in France 😉

IN THE NEWS

• OnlyFans reveals impressive financials, with $6.3 billion in revenue in 2024 and sustained growth despite its current size.

• The emergence of consciousness in AI is prompting reflection on the ethical and societal implications of machines potentially able to think and feel like us.

• Adobe plans to bring video generation to its Firefly AI model this year, with features like Generative Extend and Text to Video, while putting guardrails in place to govern content creation.

• Statistics show that AI has generated more than 15 billion images since 2022, with an average of 34 million images created every day, outpacing the speed and scale of traditional photography’s development.

YOUR DAILY DOSE OF CONTENT

1/ 🎥 Sora videos
OpenAI unveils new Sora videos made by professionals, showing striking realism that comes close to human work.

David Sheldrick · Sora Showcase

2/ 🧠 Intelligence test
Futurist Daniel Jeffries put ChatGPT through his personal IQ test, with never-before-seen questions, and it got a perfect 100% score.

o1-preview is incredible so far.  Absolutely incredible.

I am running it through my private intelligence test which I call the AIQ test. I pulled many of the problems from old, out of print intelligence test and math problem books and I wrote my own variations once I learned the… x.com/i/web/status/1…

7:37 PM • Sep 12, 2024

254 Likes   33 Retweets  

THAT’S ALL FOR TODAY

Thanks for reading the work of our team of purr-sistent writers 😼

💡 To help me improve, develop new ideas or partnerships → [email protected] or @mattioo81

👍️ Like what you’re reading? Subscribe!

📺 If you prefer video content, I invite you to subscribe to my YouTube channel.

– Mathieu