Former Google AI researcher opens an AGI Lab

Hello AI enthusiasts! Here’s what you need to know about AI today:

👉 Former Google AI researcher opens an AGI Lab

👉 MatterGen revolutionizes material design with AI

👉 Apple stops its AI-generated faulty notifications

and many more!

📧 Did someone forward you this email? Subscribe here for free to get the latest AI news everyday!

Read time: 4.8 minutes

What’s going on: François Chollet, an influential AI researcher and the creator of the Keras deep learning library, has co-founded Ndea, a new startup dedicated to advancing Artificial General Intelligence (AGI). The company will focus on developing AI systems that can learn, adapt, and innovate autonomously, using a technique called “program synthesis”.

What does it mean: The establishment of Ndea is a significant push towards achieving AGI, which is the next step for AI. It shows a shift in AI research towards more ambitious goals beyond narrow AI applications. Before this, we only had OpenAI seriously rooting for building AGI but now, there is a strong competitor in the scene!

More details:

• François Chollet is a former Google AI researcher and also the creator of the famous ARC-AGI benchmark, on which OpenAI’s latest o3 model scored 87.5% and turned a lot of heads.

• Chollet is not alone. Mike Knoop, co-founder of Zapier is his co-founder and thay are assembling a team focused on deep learning-guided program synthesis.

• Program synthesis is a technique that allows AI to generalize problems it hasn’t seen before from only a few examples. Despite being computationally expensive, Chollet and Mike are confident that they can overcome this problem. (Watch this video to learn more)

• Ndea is actively recruiting for research positions. (In case you want to join)

What’s going on: Microsoft researchers have introduced MatterGen, a generative AI model that revolutionizes materials discovery in science by directly generating novel materials tailored to specific needs, such as efficient solar cells or materials for CO2 recycling. This tool marks a significant shift from traditional methods that relied on screening existing materials to a process where AI proposes new compounds based on desired properties.

What does it mean: The introduction of MatterGen represents a paradigm shift in materials science, potentially accelerating the development of new technologies for renewable energy, sustainable materials, and environmental applications. By using AI to directly generate materials with precise properties, it bypasses the slower, more resource-intensive trial-and-error approaches, thereby speeding up innovation cycles.

More details: 

• This could lead to faster advancements in fields like battery technology.

• Alongside MatterGen, Microsoft has developed MatterSim, a complementary AI tool that simulates material properties.

• The tool generates thousands of candidates with user-defined constraints to propose new materials that meet specific needs.

• The fast progress of MatterGen even surprised its creators! Ziheng Lu, a key innovator behind this tool says: “We thought it might be possible for machine-learning-based methods to replace 80–90% of quantum-mechanical calculations within one or two years, but progress happened so quickly that the tools began outperforming our initial expectations”.

• Microsoft has also made MatterGen open-source (GitHub repository)

✋🏼 Apple has temporarily suspended its AI-generated news notification summaries due to inaccuracies in the alerts produced.

🏍 Intelligent Cranium Helmets has introduced the iC-R, an AI-enhanced motorcycle helmet that offers 300-degree visibility, crash detection, and seamless connectivity.

🤖 Italian AI startup iGenius has launched Colosseum 355B, a large language model for highly regulated industries and is expecting $1 billion for completing a data center project with Nvidia.

⚖ Legal AI startup Harvey is in talks for a $300 million funding round led by Sequoia Capital, potentially doubling its valuation to $3 billion.

📽 Luma Labs has released Ray2, a new generative video model that enhances motion fluidity and physics accuracy for more realistic video production.

💎 Google's Gemini holds a 3.1% market share in paid AI tools, significantly behind ChatGPT's 62.5%, but Google has plans for Gemini to reach 500 million users by the end of 2025.

Writing code with AI

Sometimes you might need a little bit of inspiration while coding or you might be looking for new ideas, or implementing features faster. In those cases, you can provide context of your software and ask AI directly for creating functions/classes you need.

BrightAI - Frontend Engineer

VISA - Data Scientist

Flagship Pioneering - Senior Machine Learning Ops Engineer

Vacasa - Sr Data Scientist

Thank you for staying with us like always! If you are not subscribed, subscribe here for free to get more of these emails in your inbox! Cheers!