🤖 China flexes its muscles in the humanoid robot race

Hey AI team! We’re back with a new edition of Five Minute AI this Friday ⏰

On today’s menu:

China shows off 27 humanoid robots at the 2024 World Robot Conference in Beijing.
CodeSignal offers an AI conversation simulator for practicing all kinds of work situations.
Midjourney unveils its new web interface for AI image creation.
Researchers explore the idea of “personhood credentials” to tell humans apart from AIs online.

Ready? Let’s go...

THE BIG STORY

🤖 China takes on Tesla in humanoid robots

The 2024 World Robot Conference in Beijing was the stage for an impressive show of force. No fewer than 27 Chinese humanoid robots were on display alongside Tesla’s Optimus, the only foreign entry. This massive showcase reflects China’s ambition to become the leader in this fast-growing field.

Details:

Among the standout innovations, AGIBOT, founded by a Huawei alum, unveiled robots powered by advanced language models, aimed at industry and customer service.
Other notable creations include Astribot’s S1 assistant robot, which can do calligraphy and play musical instruments, and Galbot’s wheeled robots designed for meal delivery and retail tasks.

Why it matters:

While China is still seen as slightly behind the United States in the AI race, this event clearly shows its determination to take the lead in humanoid robots. The range of applications on display hints at a future where these robotic assistants could become a commercial reality sooner than expected. That said, it’s worth noting that technological and economic challenges remain. The next few years will be crucial to see how this competition drives innovation and shapes the way we’ll interact with humanoid robots.

🗣️ CodeSignal’s AI conversation simulator: sharpen your communication skills

In a constantly changing work world, communication is an essential skill. CodeSignal recently launched an innovative tool to help you polish your communication abilities: an AI conversation simulator. It lets you practice all kinds of professional situations in a safe, controlled environment.

Details:

Pick from a variety of professional scenarios, such as job interviews, team management discussions or conversations with clients.
Take part in a full voice conversation with an AI partner, for an experience that feels close to the real thing.
Get instant, personalized feedback from Cosmo, your AI tutor, to pinpoint your strengths and areas for improvement.

Why it matters:

CodeSignal’s simulator is an interesting opportunity to build your communication skills in a practical, effective way. Whether you’re getting ready for a big interview or want to improve your everyday interactions at work, this tool can help you gain confidence and ease. Give it a try and see for yourself how it can help your communication skills grow.

LEARN TO MASTER AI

🎨 Midjourney on the web: a new era for AI image creation

I mentioned it briefly in previous newsletters, and today I decided to put together a tutorial on using Midjourney with the new web interface.

This change opens up new possibilities for creators and anyone curious about generative AI.

Here’s how to start your Midjourney adventure in a few simple steps:

Sign up on the official Midjourney website with your Gmail account.
Choose the plan that works for you, starting at $10 a month.
Head to the “Create” page to launch your first generations.
Here’s a video tutorial in French to get familiar with the interface and discover lots of ways to use it. I’m also including the “official” tutorial that Midjourney posted on X.

To get the most out of your creations, here are a few practical tips:

Write your prompts in English, so the AI understands them better.
Explore the image format options (square by default) and the stylization options in the settings. “Stylization” boosts Midjourney’s signature look, since it’s trained to deliver artistic compositions and contrasting colors. Above 400, the generator starts taking a lot of liberties with your prompt.
Use this prompt generator to help you structure your prompts.
Try the “–q 2” parameter to improve the quality of your renders (watch your credit usage).

Midjourney on the web makes this powerful visual creation tool easier to access. Whether you’re an artist, a designer or just curious, now is the perfect time to explore what this constantly evolving technology can do.

WHAT THE RESEARCH SAYS

🧠 A proof of humanity to browse the internet?

A team of researchers from OpenAI, Microsoft, MIT and other prestigious institutions is working on a novel concept: “Personhood Credentials” (PHC), or “certificats d’humanité” in French. The initiative aims to tackle one of the major challenges of our digital age: telling humans apart from AI bots online.

Details:

PHCs would be digital credentials certifying that their holder is human, without revealing any other personal information.
The system would exploit two current limitations of AI: its inability to perfectly imitate humans offline and to get around modern cryptographic systems.

Why it matters:

As online interactions grow more and more complex, this research opens up interesting avenues for preserving the authenticity of exchanges on the internet. PHCs could strike a balance between privacy protection and fraud prevention, while letting us navigate a constantly changing digital world with peace of mind. Although the concept raises ethical and practical questions, it deserves our attention because it could shape the future of our online interactions.

🧩 AI learns to plan better… all on its own!

IBM Research and Cornell University just made an exciting breakthrough with the creation of AutoToS, a system that teaches AI how to solve complex planning problems with 100% accuracy, and without any human intervention!

Details:

AutoToS acts as a smart tutor for AI, helping it break down and solve complex problems step by step.
The system uses clever tests to check the AI’s work, pointing out its mistakes and showing examples of how to improve, all autonomously.
The approach seems to work just as well for small AI models as for large ones.

Why it matters:

AutoToS marks an important step toward more reliable and autonomous AI assistants. While it’s currently hard to fully trust AI agents with complex tasks, this system paves the way for concrete, precise applications. Imagine a virtual assistant that can plan your day or solve complex logistics problems without ever making a mistake! We’re not there yet, but AutoToS is a promising step toward more capable and trustworthy AI in our daily lives.

THE LATEST TRENDING TOOLS

💡 OTHER TOOLS:

🎭 Adobe Express Animate from Audio turns your audio files into character animations with AI, adding a new dimension to your sound creations.
📚 Paperguide makes it easier to manage and understand research papers, making academic exploration more accessible and efficient.
💡 Kraftful 3.0 helps you build product strategies with AI-generated insights, offering a fresh perspective on product development.
🖼️ Finegrain Image Enhancer improves the quality of low-resolution images, giving your visuals a second life.
⏰ Reclaim offers an AI calendar that smartly balances work and personal life, optimizing your time management.
🗣️ D-ID Video Translate translates your videos into multiple languages, clones the speaker’s voice and perfectly syncs the lips in a single click, making it easier to share multilingual content.

IN THE NEWS

Google is exploring the use of transformers to improve music recommendations, opening new possibilities for personalizing the listening experience.
Tesla is hiring people to train its Optimus robot through motion capture, reaching a new milestone in humanoid robot development.
ElevenLabs is launching its AI text-to-speech app Reader worldwide, supporting more than 30 languages and hundreds of new voices.
Google unveils a prompt gallery for AI Studio, giving developers 31 prebuilt templates covering a wide range of applications.
Mark Zuckerberg and Daniel Ek make the case for open-source AI adoption in Europe, warning against overly complex regulations that could stifle innovation.
Fetch.ai opens its Innovation Lab in San Francisco, along with a $10 million fund to support early-stage AI agent startups.

YOUR DAILY CONTENT FIX

1/ 🤖 AI avatar: Nvidia pushes the limits of real-time interaction

Check out Nvidia’s AI avatar, which lets you interact with it directly for 3 minutes, a promising experience even if it still has room for improvement.