Google DeepMind, the company's artificial intelligence research lab, has a new molecular prediction model called AlphaFold 3 that predicts the interactions of “all the molecules of life” with at least 50% greater accuracy than past models. It is said to be predictable. This means scientists can develop drugs that more effectively target diseases, Google said in a blog post and a paper published Wednesday in the journal Nature.
AlphaFold 3, which is freely available to researchers, allows scientists to test different structures made of proteins, DNA, and RNA to predict how different materials will interact.
Performing this type of predictive modeling on AlphaFold 3 has the potential to significantly speed up research while reducing costs. According to Google, creating experimental protein structure predictions previously took years and cost hundreds of thousands of dollars.
Google acquired DeepMind in 2014 for an estimated $400 million to $650 million to expand its AI capabilities. DeepMind has made headlines for developing a system that can beat the world's best Go, Chess, and Shogi players. Immediately, advances began in the discovery of protein folding and crystal structures.
AlphaFold 2 has been used not only for enzyme design but also for malaria and cancer treatment, Google said. AlphaFold 3 takes things a step further by also focusing on a “broad range of biomolecules,” which could be useful for everything from biorenewable materials discovery to drug design and genomics research.
Google said it has already been used to predict hundreds of millions of structures that would “take researchers hundreds of millions of years at the rate of current structural biology experiments.”
DeepMind's sister company, Isomorphic Labs, is already working with pharmaceutical companies to develop new treatments.
DeepMind's advances in supporting scientific research are not only beneficial to humanity, but also give Google a positive brand image as a technology company that wants to use its power and expertise to help solve critical challenges. is also helpful. DeepMind's advances will also help Google enter new markets.
To learn more about Google's everyday AI tools, be sure to read CNET's hands-on reviews of Gemini and ImageFX, as well as reviews of chatbots like Claude, Adobe Firefly, and OpenAI's ChatGPT.
Editor's note: CNET used an AI engine to create dozens of stories and label them accordingly. The notes you are reading are attached to articles that substantively cover the topic of AI, all written by professional editors and writers. Please see our AI Policy for more information.