Key insights

• ⚙️ Simulating 500 million years of evolution using ESM3 language model
• 🏎️ Etched releasing the fastest AI chip with support from various big names
• 🧬 AlphaFold's ability to predict complex 3D shapes of proteins
• ⚕️ Proteins as programmable building blocks of life and their potential for drug development
• 🧬 AI mimicking natural evolution to create a new glowing protein
• ⚛️ AI model generates proteins across sequence, structure, and function modalities
• 🔬 Using evolutionary principles and natural proteins to create new glowing proteins
• 📊 ESM is an open-source model for protein programming with 1.4 billion parameters

Q&A

• What are some recent developments in AI chip technology and protein research?

  Recent developments include the release of the fastest AI chip by the company etched, Asic chips specialized for running Transformer models, and evolutionary scale.ai's research on protein creation and carbon capture, claiming to outperform Nvidia's GPUs.

• What is the ESM, and how is it relevant to protein programming?

  The ESM is an open-source model for protein programming with 1.4 billion parameters, available on GitHub for non-commercial use, written in Python. The model is being released by a startup formed by former members of The Meta Fair protein group with ties to Facebook/Meta.

• What techniques were used to create new glowing proteins, and what were the key outcomes?

  Scientists used evolutionary principles, natural proteins, and machine learning techniques to create new glowing proteins that differ from natural ones, with the AI model generating 96 protein candidates, including a non-natural, very bright one called C10 using a Chain of Thought approach similar to Chad GPT for reasoning.

• How does the AI model contribute to generating proteins and their significance?

  The AI model can generate proteins across sequence, structure, and function modalities, providing unprecedented control for scientists to prompt new protein designs, achieve atomic-level accuracy, develop new skills at scale, and provide self-feedback. This advancement is significant in creating life forms to break down plastic waste.

• What are some exciting advancements in AI related to simulating evolution and biological engineering?

  Exciting advancements include the use of AI to simulate 500 million years of evolution using the ESM3 language model, creating a new glowing protein through the mimicking of natural evolution, and the AI's potential in biological engineering and protein analysis.

• 00:00 Exciting advancements in AI: simulating 500 million years of evolution, fastest AI chip by etched, and programmable biology using language models. Big potential for drug development and understanding proteins.
• 02:40 Scientists have used AI to create a new glowing protein that mimics natural evolution, demonstrating the potential of AI in biological engineering.
• 05:25 An AI model can generate proteins across sequence, structure, and function modalities, offering unprecedented control. It can prompt new protein designs, achieve atomic-level accuracy, develop new skills at scale, and provide self-feedback. This advancement is significant in creating life forms to break down plastic waste.
• 08:02 Scientists have used evolutionary principles, natural proteins, and machine learning to create new glowing proteins that differ from natural ones and are 20% different. The AI model uses a Chain of Thought approach, similar to Chad GPT, to reason step by step. It generated 96 protein candidates, including a non-natural, very bright one called C10.
• 10:44 The ESM is an open-source model for protein programming with 1.4 billion parameters, and the largest model is 98 billion parameters. It's available on GitHub for non-commercial use with the code written in Python. The model is being released by a startup formed by former members of The Meta Fair protein group with ties to Facebook/Meta.
• 13:33 Exciting developments in AI chip technology and evolutionary scale.ai's research on protein creation and carbon capture. Asic chips specialized for running Transformer models claim to outperform Nvidia's GPUs.