- The 79
- Posts
- Google’s AI makes scientific discoveries without humans
Google’s AI makes scientific discoveries without humans
Hi everyone! Here’s what you need to know about AI today:
👉 Google’s new AI co-scientist can make discoveries on its own
👉 Nvidia just dropped a massive biology AI model
👉 Google is removing Gemini from the main Google app on iOS
and many more!
📧 Did someone forward you this email? Subscribe here for free to get the latest AI news everyday!
Read time: 5 minutes
Google is developing a ground-breaking tool for scientists
Source: Google Research | AI co-scientist
What’s going on: Google Research announced the development of an AI co-scientist designed to accelerate scientific breakthroughs. This innovative tool aims to assist researchers by using AI to tackle complex problems across various scientific domains. The AI co-scientist acts as a collaborative partner that can process vast amounts of data, generate hypotheses, and even suggest experimental designs.
What does it mean: AI co-scientist integrates advanced machine learning techniques, allowing it to learn from existing scientific literature and experimental results. By doing so, it can identify patterns or insights that might elude human researchers due to the sheer volume of information involved. Google emphasizes that this AI is not just a passive tool but an active participant in the scientific process, capable of proposing novel research directions and refining its understanding as new data becomes available.
More details:
The AI co-scientist uses test-time compute scaling to iteratively reason, evolve, and improve outputs. Key reasoning steps include self-play-based scientific debate for novel hypothesis generation, ranking tournaments for hypothesis comparison, and an "evolution" process for quality improvement.
This tool can accelerate discoveries in fields like biology, chemistry, and physics.
While performing validation tests, AI co-scientist proposed novel repurposing candidates for acute myeloid leukemia (AML), identified epigenetic targets for liver fibrosis, and shed light on mechanisms of antimicrobial resistance.
Interested research organizations around the world can be an early-tester of AI co-scientist by joining Google’s Trusted Tester Program.
NVIDIA
Nvidia and Arc released the largest biology AI model in the world
Source: Nvidia Blog
What’s going on: In collaboration with Arc Institute and Stanford University, Nvidia introduced Evo 2, a powerful new model in biomolecular AI designed to accelerate scientific discovery in fields like drug development and protein engineering. Building on its predecessor, Evo 1, this next-generation model leverages Nvidia’s expertise in AI and GPU computing to tackle complex biological challenges at an unprecedented scale. The system, trained on an enormous dataset of nearly 9 trillion nucleotides, is engineered to simulate and predict biomolecular interactions with remarkable precision, offering researchers a powerful tool to explore the intricacies of life at the molecular level.
What does it mean: Evo 2 can process vast datasets and perform intricate simulations that were previously computationally infeasible. By integrating generative AI techniques, the model can design novel proteins and optimize molecular structures. Evo 2’s architecture takes advantage of GPU-enabled parallel processing that dramatically speeds up tasks like protein folding predictions and ligand binding analyses.
More details:
In drug discovery, Evo 2 can identify drug candidates by simulating how molecules interact with biological targets. In healthcare, Evo 2 could help researchers understand which gene variants are tied to a specific disease.
The NVIDIA NIM microservice for Evo 2 allows users to generate a variety of biological sequences, with settings to adjust model parameters. Developers are also able to fine-tune Evo 2 on their proprietary datasets by downloading the model through the open-source Nvidia BioNeMo Framework.
Evo 2 was built using Nvidia DGX Cloud on Amazon Web Services (AWS) and is available to developers worldwide on the Nvidia BioNeMo platform.
👨💻 Codeium, AI coding startup, is in discussions to raise funding at a valuation nearing $3 billion.
☄ Karman+, a startup focused on asteroid mining, has raised $20 million in seed funding to develop an autonomous spacecraft capable of extracting materials like water from asteroids.
📱 Google is removing its AI assistant Gemini from the main Google app on iOS to encourage users to download the standalone Gemini app.
📞 Sanas, AI speech-to-speech startup, has raised $65 million in funding to expand its AI technology that modifies call center workers' accents in real-time.
📈 Mistral's AI assistant, Le Chat, achieved over one million downloads in just 14 days after its release, securing the top spot for free downloads on the iOS App Store in France.
🦠 Google's new “co-scientist” AI solved a decade-long superbug mystery in just two days, identifying how certain antibiotic-resistant bacteria use viral tails to spread across species.
AI + Well-being
I want you to help me create a weekly fitness plan. Think about a time when I might have felt strong, energized, and at peace, maybe after a workout, a walk, or a calm moment. Identify key elements that could have made that possible (like nutrition, movement, or rest), and suggest a sustainable plan that incorporates one small, actionable change I can start tomorrow. Provide specific ideas for activities, timing, and how it supports my well-being.
Grok 3’s answer
Anduril - Senior Machine Learning Engineer
Chewy - Machine Learning Engineer II
NBCUniversal - Staff Deep Learning Engineer
Thank you for staying with us like always! If you are not subscribed, subscribe here for free to get more of these emails in your inbox! Cheers!