AlphaFold 3 Predicts All Molecules

PLUS: OpenAI Unveils the Model Spec

Hello readers,

Welcome to another edition of This Week in the Future! Google DeepMind released AlphaFold 3, which predicts the interactions of all of life’s molecules, and OpenAI asked for public feedback in shaping its future AI models.

Let’s get into it!

AlphaFold 3

Google DeepMind and Isomorphic Labs have released AlphaFold 3, a new AI model that not only predicts the structure of proteins, but the structure and interactions of all of life’s molecules. The goal is to transform biology and drug discovery.

For interactions between proteins and other molecule types, AlphaFold 3 has 50% higher prediction accuracy than traditional methods. Scientists will be able to access the model for free through the newly launched AlphaFold Server.

Why This Matters

Isomorphic Labs is already collaborating with pharmaceutical companies to apply AlphaFold 3 in drug design. If we can solve biology, then there might be no limit to how well we can improve our health and live forever!

OpenAI Opens the Floor

OpenAI has published multiple explanations of itself starting with the Model Spec, where they tell us how they shape desired model behavior. The first draft of the Model Spec is intended to be a living document. OpenAI is inviting public feedback, which can be submitted here. You can help shape future AI models or at least feel like you did.

Additionally, OpenAI explained their approach to training data, a point of contention for artists and publishers. OpenAI says that “learning is fair use,” an interesting use of language that will probably catch on. However, OpenAI seems to know in their hearts that something is ethically amiss, so they are allowing content owners to opt out of training and will launch a new Media Manager in 2025 (after they’ve scraped everything).

Also from OpenAI:

Our Take

OpenAI’s efforts are commendable, but they also suggest some insecurity about the more ethically dubious aspects of current AI. The question OpenAI seems to be grappling with is: if one’s end is to build a superintelligence that will propel humanity into the future, then what means wouldn’t be justified?

🔥 Rapid Fire

📖 What We’re Reading

“In an era where generative AI and advanced technologies are reshaping industries, 2024 is emerging as a pivotal year in the evolution of B2B ecommerce. The B2B Ecommerce Site Search Trends 2024 report, a study Algolia recently conducted in partnership with Escalent, captures this transformation, focusing on how B2B organizations are leveraging AI-driven search and discovery platforms to enhance their digital capabilities and drive sales.”

Source: AI For All

“Asset managers are facing a variety of challenges to their growth. Investors are gravitating to passively managed funds and other products that have lower fees even as asset managers’ costs increase. As the AI technological revolution gathers momentum, asset managers have an opportunity to invest in AI and integrate it into their operations in ways that can enhance a three Ps strategy.”

Source: Boston Consulting Group

💻️ AI Tools and Platforms

  • Dataminr → Real-time generative AI for event and risk detection

  • Qdrant → Vector database for next-generation AI applications

  • BigPanda → AI-powered ITOps and incident intelligence

  • GovDash → AI platform for government contractors

  • AI Squared → Render AI insights directly into business apps

Learn AI in 5 Minutes a Day

AI Tool Report is one of the fastest-growing and most respected newsletters in the world, with over 550,000 readers from companies like OpenAI, Nvidia, Meta, Microsoft, and more.

Our research team spends hundreds of hours a week summarizing the latest news, and finding you the best opportunities to save time and earn more using AI.