Key Takeaways –
-
AlphaGenome offers researchers a fresh method to examine the impacts of billions of potential DNA variants.
-
Co-Scientist leverages Gemini-powered AI agents to generate, evaluate, and refine scientific hypotheses.
-
Hassabis has transitioned into a broader science-focused role at Alphabet as Isomorphic Labs continues to scale its AI-driven drug discovery initiatives.
Demis Hassabis has assumed a new position within Alphabet while maintaining his focus on leveraging advanced artificial intelligence to tackle complex scientific hurdles. On August 5, 2026, Hassabis was named Chair of Google DeepMind and Chief Scientist of Alphabet. Meanwhile, Koray Kavukcuoglu took over day-to-day operations at Google DeepMind alongside the advancement of Gemini. Hassabis additionally keeps his role as CEO of Isomorphic Labs, where AI plays a direct part in finding new drugs.
AlphaFold Set the Foundation for AI Science
AlphaFold remains the most prominent demonstration of Hassabis’s science-centric strategy. According to Google, the AlphaFold database has been accessed by upwards of 3 million researchers spread across more than 190 countries. The repository currently houses over 200 million predicted protein structures. Research connected to AlphaFold has yielded more than 35,000 citations, while AlphaFold 2 alone has been referenced in over 200,000 papers.
AlphaFold 2 successfully tackled a half-century-old challenge regarding protein structures. This achievement earned Hassabis and John Jumper the 2024 Nobel Prize in Chemistry. That triumph naturally prompted a bigger question: Can artificial intelligence help transition science from mere prediction to active discovery?
AlphaGenome Maps DNA Changes
AlphaGenome applies this foundational concept directly to the field of genetics. On September 8, 2026, Google DeepMind unveiled the AlphaGenome Atlas, a database featuring predictions for roughly 9 billion individual-letter DNA variants found in the human genome. Consuming approximately 1 petabyte, the dataset is over 30 times larger than the AlphaFold database.
By helping to narrow down a massive search space, AlphaGenome utilizes its AlphaGenome Variant Impact score to enable scientists to rank variants based on their potential biological outcomes. A project conducted by the University of Exeter paired the AlphaGenome Atlas with health information from over 54,000 UK Biobank participants. Through this analysis, a specific region containing 526 candidate variants was successfully winnowed down to just four, providing a much clearer route for biological research.
Also Read – Meta Muse AI: 15 Useful Prompts & Use Cases You Should Try
Co-Scientist Gives AI a Role in Hypothesis Work
Google DeepMind rolled out Co-Scientist in May 2026. The platform utilizes Gemini alongside a suite of AI agents to formulate, test, critique, and improve scientific hypotheses. DeepMind reports that over 100 institutions have already experimented with the technology.
Initial investigations have spanned antimicrobial resistance, plant immunity, liver fibrosis, ALS, infectious diseases, and molecular biology. Regarding the liver fibrosis case specifically, DeepMind notes that Co-Scientist identified drug-repurposing candidates that had previously been missed by researchers. One such candidate successfully blocked 91% of a fibrosis-associated reaction during laboratory trials.
Outputs of this nature still demand rigorous validation. A compelling AI hypothesis is not the same as a verified treatment, and turning any concept into a safe medicine requires extensive lab work, clinical trials, and regulatory oversight.
Gemini Connects Scientific Tools
Google’s current objective is to turn Gemini into a comprehensive scientific workbench. Gemini for Science integrates Co-Scientist alongside platforms like AlphaGenome, AlphaFold, and AlphaEvolve, while also granting researchers access to more than 30 life-science tools and databases.
Rather than deploying a single model for a single isolated chore, Google’s vision is centered on a generalized AI framework capable of orchestrating specialized models, research databases, and technical tools. WeatherNext 3, which debuted on September 3, 2026, applies this identical strategy to meteorology. Google states that this system provides precipitation forecasts that are 50% more accurate one or more days in advance compared to its previous iteration.
AI Moves into Physical Experiments
In December 2025, Google DeepMind revealed blueprints for its inaugural automated science laboratory in the UK, with an initial emphasis on materials science. The facility is designed to pair Gemini with robotic systems capable of synthesizing and testing hundreds of distinct materials daily.
This configuration establishes a tighter feedback loop for research: an AI suggests a concept, a robot executes the physical test, laboratory instruments record the outcomes, and the resulting data informs the next wave of hypotheses.
Also Read – CRO 2027 Checklist: 10 Revenue Priorities for AI-Driven Growth
Isomorphic Labs Pushes AI into Drug Discovery
Hassabis’s endeavors through Isomorphic Labs introduce another critical pillar to this overarching strategy. In May 2026, Isomorphic Labs secured $2.1 billion in Series B funding, intending to scale its computational drug-design platform and advance its therapeutic pipeline toward clinical trials. Key industry partners include Novartis, Eli Lilly, and Johnson & Johnson.
Ultimately, this expansive vision resembles less of a standalone breakthrough model and more of an integrated scientific engine. AlphaFold maps proteins, AlphaGenome evaluates DNA variants, and Co-Scientist generates hypotheses. Meanwhile, Gemini ties these computational resources to broader research databases, and automated laboratories put these theories to the test in the physical world.
Hassabis’s recent promotion at Alphabet grants this grand vision an even wider platform, cementing his role right at the crossroads of frontier AI, medical research, and fundamental science. Enhanced artificial intelligence can assist researchers in navigating exploration spaces that human effort alone could never cover at an equivalent scale.
FAQs
1. Who is Demis Hassabis?
Demis Hassabis is a co-founder of Google DeepMind and a prominent figure in artificial intelligence research and scientific innovation.
2. What is AlphaFold?
AlphaFold is an artificial intelligence model designed to predict protein structures, serving as an essential asset for biological investigations.
3. What does AlphaGenome do?
AlphaGenome forecasts the consequences of DNA alterations, assisting scientists in pinpointing genetic variants carrying significant biological implications.
4. What is Co-Scientist?
Co-Scientist is a Gemini-powered framework built to formulate, test, critique, and refine scientific hypotheses across diverse academic fields.
5. How does Isomorphic Labs fit into Hassabis’s work?
Isomorphic Labs applies artificial intelligence to pharmaceutical research, utilizing advanced computational approaches to help engineer novel medicines.
