Welcome to this new edition!

In Today’s Menu:

• AI-related Quote

• Recommendation of the week

• AI-ducation: How AI learns from Mistakes

• AI decrypted: Generative Adversarial Networks (GANs)

• Top 3 News of the Week

• AI story: Andrew Ng

• Extra News

• Image of The Day

All your news. None of the bias.

Be the smartest person in the room by reading 1440! Dive into 1440, where 3.5 million readers find their daily, fact-based news fix. We navigate through 100+ sources to deliver a comprehensive roundup from every corner of the internet – politics, global events, business, and culture, all in a quick, 5-minute newsletter. It's completely free and devoid of bias or political influence, ensuring you get the facts straight.

Subscribe to 1440 today.

The AI-related Quote

Recommendation of the week

I was personally able to chat with Arxiv's founders via The AI Newsroom LinkedIn account. Their commitment to making quality content led me to the point of accepting a partnership with them, given that we share this crucial point for me.

By supporting them, you support me! ❤️

The AI-ducation

How AI Learns from Mistakes

AI is like a student. It learns by making mistakes and improving. When you train an AI, it starts by making lots of errors. For example, a self-driving car AI might not know when to stop or turn. It keeps making these mistakes, but each time it does, it learns something new.

This process is called "machine learning." The AI gets better by comparing its results to the correct answers. It adjusts, tries again, and improves over time. Just like humans, it needs a lot of practice to become good at something.

The more data it gets, the smarter it becomes. It’s like giving a student more books to read. The more information the AI receives, the better it gets at making decisions. And just like that, AI keeps learning, becoming smarter every day.

The AI Decrypted

What are Generative Adversarial Networks (GANs)?

Generative Adversarial Networks, or GANs, are a type of AI that can create new things. Think of it as two AIs working together like a team. One AI (called the Generator) tries to create something new, like a picture or a sound. The other AI (called the Discriminator) checks if what the Generator made looks real or fake.

The two AIs play a game. The Generator wants to fool the Discriminator by creating things that look real. The Discriminator's job is to catch the fakes. Over time, both AIs get better. The Generator learns to make more realistic creations, and the Discriminator becomes better at spotting fakes.

This back-and-forth process is how GANs create things like realistic images, art, or even music. It's a powerful tool that helps AI create new, never-before-seen content.

The 3 News of the Week

#1 - A More Efficient AI: OLMoE

Researchers from the Allen Institute for AI, Contextual AI, University of Washington, and Princeton University introduced a new model called OLMoE. This open-source model is faster and cheaper to run than large models like GPT-4 and Llama2.

What makes OLMoE special is its efficiency. It only uses 1 billion out of its 7 billion parameters at a time, which cuts down on resources. In fact, OLMoE is 10x more efficient than models like Llama2-13B and GPT-4, without sacrificing performance.

#2 - New 8B Model Outperforms LLaMA-3.1-8B

AntA new 8-billion parameter language model is making waves by outperforming Meta’s LLaMA-3.1-8B-Instruct and Hermes-3-LLaMA-3.1-8B in 7 out of 9 benchmark tests. This model excels in tasks like answering questions, following instructions, and reducing mistakes.

The key to its success? The team used advanced techniques like self-curation and fine-tuning to help the model learn from the best examples. This makes it work efficiently, even on limited hardware. They also used a technique called model merging, blending two strong models with SLERP to create something even better.

Thanks to these methods, this model is achieving impressive results on important AI benchmarks.

#3 - UN Chief Calls for Global Cooperation in AI

UN Secretary-General António Guterres, speaking at an AI workshop in Shanghai co-hosted by the UN and China, called for international cooperation to ensure AI benefits everyone. He emphasized that AI’s potential should be shared equally, warning that unequal access could worsen global inequalities.

Guterres highlighted the concentration of AI tools in a few companies and countries, leaving many nations behind. He stressed that bridging this gap is critical, especially with AI's ability to support Sustainable Development Goals (SDGs). International cooperation and solidarity, he said, are key to unlocking AI's full potential.

He also warned that AI risks, like its benefits, are unevenly distributed. Without proper regulations, AI could widen digital divides and disproportionately impact vulnerable populations. The upcoming Summit of the Future is seen as a crucial moment for establishing inclusive AI governance.

The AI Story

Andrew Ng

Andrew Ng is one of the most influential minds in AI today. He co-founded Google Brain, where he worked on deep learning projects that transformed the AI landscape. His research helped lay the foundation for the rise of neural networks in AI applications, making them more accessible and powerful.

Ng is also known for making AI education available to the masses. Through his popular online courses on platforms like Coursera, he has taught millions about machine learning and AI, helping to democratize knowledge in this field. His efforts have inspired a new generation of AI researchers and engineers.

In addition to his academic contributions, Ng continues to push for AI development that benefits everyone. As the co-founder of Landing AI, he is focused on helping industries outside of tech use AI to solve real-world problems, from manufacturing to healthcare. His work is shaping the future of AI in practical, impactful ways.

The Extra News

• Elon Musk denies Tesla will pay xAI for AI technology partnership → here

• Senator David Pocock creates AI deepfakes of Anthony Albanese and Peter Dutton to call for ban ahead of election → here

• UK signs first international treaty to implement AI safeguards → here

Image of The Day

Prompt: 
A cute kitten looking in the mirror and seeing an elegant lion reflection, high resolution, highly detailed, photorealistic, cinematic —ar 3:4 —style raw

Thank you for your time!

See you next Tuesday at 9:12am!

Hey readers, Doug here!

I'd like to sincerely thank you for taking the time to read The AI Newsroom every week.

I make every effort to send you valuable emails every week.

Please let me know how you found this edition by replying to this email or by answering the questionnaire below. 👇

♻️ Please also feel free to share as much of the newsletter as you can with your friends, colleagues, or AI girlfriend. It helps me a lot!

Keep Learning!