We couldn't have achieved this without the foundational work of the structural biology community, which provided the initial 150,000 protein structures that our system learned from.

We released the protein structures to the world knowing that the scientific community would find countless applications for them, far beyond what we could imagine.

Seeing the map of the globe light up as researchers worldwide accessed the protein structures we released was a humbling and amazing experience.

There's enormous potential in specializing general techniques for specific applications, such as protein structure prediction, and we are actively exploring this in our AI for science work.

AlphaFold 3 represents a significant breakthrough in understanding biomolecular structure, allowing for experimental-level accuracy in predicting protein structures and their interactions with small molecules, DNA, and RNA.

More than 2 million researchers worldwide now use AlphaFold, and it has had a significant impact on biology and medical research, but this is just the beginning of AI's potential in science.

The future may involve programming proteins as a new form of technology, bridging the gap between computers and human health, which is currently underexploited. With advancements in machine learning, like AlphaFold, we can now predict how DNA sequences fold into proteins in just minutes, potentially revolutionizing how we build machines that interact with the human body and our environment.

AlphaFold 3 is a significant step towards solving the structure prediction problem, but we still aim for 100% accuracy in understanding molecular interactions.

AlphaFold has become a standard tool for biology and medical research, with over 2.5 million researchers worldwide using it in their critical work, significantly impacting the scientific community.