Introducing AlphaQubit: our AI-based system that can more accurately identify errors inside quantum computers. 🖥️⚡

This research is a joint venture with @GoogleQuantumAI, published today in @Nature → goo.gle/3ZflWMn The possibilities in quantum computing are compelling. ♾️

They can solve certain problems in a few hours, which would take a classical computer billions of years. This can help lead to advances in areas like drug discovery to material design.

But building a stable quantum system is a challenge.

Our latest generative technology is now powering MusicFX DJ in @LabsDotGoogle - and we’ve also updated Music AI Sandbox, a suite of experimental music tools which can streamline creation. 🎵

This will make it easier than ever to make music in real-time with AI. ✨goo.gle/4eTg28Z MusicFX DJ lets you input multiple prompts and include details on instruments, genres and vibes to create music. 🎛️

We’ve updated and improved the interface using feedback from @YouTube’s Music AI Incubator.

We’re presenting AlphaProteo: an AI system for designing novel proteins that bind more successfully to target molecules. 🧬

It could help scientists better understand how biological systems function, save time in research, advance drug design and more. 🧵 dpmd.ai/3XuMqbX
Protein binders are promising tools in drug development and biotech.

They’ve demonstrated potential in:
🌀 binding cancer targets
🌀 blocking viral infections
🌀 modulating immune response

But traditional ways of identifying effective protein binders involve extensive lab work.

Meet our AI-powered robot that’s ready to play table tennis. 🤖🏓

It’s the first agent to achieve amateur human level performance in this sport. Here’s how it works. 🧵 Robotic table tennis has served as a benchmark for this type of research since the 1980s.

The robot has to be good at low level skills, such as returning the ball, as well as high level skills, like strategizing and long-term planning to achieve a goal.

AI systems can be powerful but opaque "black boxes" - even to researchers who train them. ⬛

Enter Gemma Scope: a set of open tools made up of sparse autoencoders to help decode the inner workings of Gemma 2 models, and better address safety issues. → dpmd.ai/gemma-scope Language models turn your text input into a series of ‘activations’ - which map the relationships between the words you’ve entered to help it write its answer. 💬

Activations at different layers in its neural network represent increasingly advanced concepts, known as ‘features’.

We’re welcoming a new 2 billion parameter model to the Gemma 2 family. 🛠️

It offers best-in-class performance for its size and can run efficiently on a wide range of hardware.

Developers can get started with 2B today → dpmd.ai/4d0MKEH We’re also introducing ShieldGemma: a series of state-of-the-art safety classifiers designed to filter harmful content. 🛡️

These target hate speech, harassment, sexually explicit material and more, both in the input and output stages.

We’re presenting the first AI to solve International Mathematical Olympiad problems at a silver medalist level.🥈

It combines AlphaProof, a new breakthrough model for formal reasoning, and AlphaGeometry 2, an improved version of our previous system. 🧵 dpmd.ai/imo-silver Our system had to solve this year's six IMO problems, involving algebra, combinatorics, geometry & number theory. We then invited mathematicians @wtgowers and Dr Joseph K Myers to oversee scoring.

It solved 4️⃣ problems to gain 28 points - equivalent to earning a silver medal. ↓

We're sharing progress on our video-to-audio (V2A) generative technology. 🎥

It can add sound to silent clips that match the acoustics of the scene, accompany on-screen action, and more.

Here are 4 examples - turn your sound on. 🧵🔊 dpmd.ai/v2a
✍️ Prompt for audio: “Wolf howling at the moon.”

Our video generation model Veo gives more control over the camera. 📹

You can prompt for:
🔘 Extreme close up
🔘 Slow-motion crane shots
🔘 Timelapses

And more. 🧵

✍️ Prompt: “Timelapse of the northern lights dancing across the Arctic sky, stars twinkling, snow-covered landscape.” ✍️ Prompt: “A panning shot of a waterfall cascading down a rocky cliff, lush greenery surrounding the falls, mist rising from the crashing water.”

Introducing Veo: our most capable generative video model. 🎥

It can create high-quality, 1080p clips that can go beyond 60 seconds.

From photorealism to surrealism and animation, it can tackle a range of cinematic styles. 🧵 #GoogleIO ✍️ Prompt: “Many spotted jellyfish pulsating under water. Their bodies are transparent and glowing in deep ocean.”

Announcing AlphaFold 3: our state-of-the-art AI model for predicting the structure and interactions of all life’s molecules. 🧬

Here’s how we built it with @IsomorphicLabs and what it means for biology. 🧵 dpmd.ai/3URDiNo AlphaFold 3 can generate the 3D structures of proteins, DNA, RNA, and smaller molecules, while also revealing how they fit together. 🧩

It can also model chemical changes to them that control the healthy functioning of cells - and when disrupted, could lead to disease.

We're announcing TacticAI: an AI assistant capable of offering insights to football experts on corner kicks. ⚽

Developed with @LFC, it can help teams sample alternative player setups to evaluate possible outcomes, and achieves state-of-the-art results. 🧵 dpmd.ai/49PGq1b 📊 Corner kicks can be challenging for AI to model due to the limited availability of data - @premierleague matches only average about 10 a game.

TacticAI uses a geometric deep learning approach to tackle this problem. → dpmd.ai/43p5Gcc

Introducing Gemini 1.5: our next-generation model with dramatically enhanced performance. It also achieves a breakthrough in long-context understanding.

The first release is 1.5 Pro, capable of processing up to 1 million tokens of information. 🧵 dpmd.ai/3SEbw4p
Gemini 1.5 was designed using a new Mixture–of-Experts (MoE) architecture, making it much more efficient to train and serve.

When tested on a set of text, code, image, audio and video evaluations, 1.5 Pro outperforms 1.0 Pro on 87% of benchmarks used for developing our LLMs.

Introducing AlphaGeometry: an AI system that solves Olympiad geometry problems at a level approaching a human gold-medalist. 📐

It was trained solely on synthetic data and marks a breakthrough for AI in mathematical reasoning. 🧵 dpmd.ai/alphageometry
AlphaGeometry is a system made up of 2️⃣ parts:
🔵 A neural language model, which can predict useful geometry constructions to solve problems
🔵 A symbolic deduction engine, which uses logical rules to deduce conclusions

Both work together to find proofs for complex geometry theorems.

How could robotics soon help us in our daily lives? 🤖

Today, we’re announcing a suite of research advances that enable robots to make decisions faster as well as better understand and navigate their environments.

Here's a snapshot of the work. 🧵 dpmd.ai/advanced-robot…
To produce truly capable robots, two fundamental challenges must be addressed:
🔘 Improving their ability to generalize their behavior to novel situations
🔘 Boosting their decision-making speed

We deliver critical improvements in both areas. ↓ dpmd.ai/advanced-robot…

Introducing FunSearch in @Nature: a method using large language models to search for new solutions in mathematics & computer science. 🔍

It pairs the creativity of an LLM with an automated evaluator to guard against hallucinations and incorrect ideas. 🧵 dpmd.ai/x-funsearch
🔎 FunSearch uses an evolutionary approach to find the “fittest” ideas, which are expressed as computer programs to be run and evaluated automatically.

An iterative procedure allows the LLM to suggest improvements to programs while the evaluator discards bad ones.

We’re excited to announce 𝗚𝗲𝗺𝗶𝗻𝗶: @Google’s largest and most capable AI model.

Built to be natively multimodal, it can understand and operate across text, code, audio, image and video - and achieves state-of-the-art performance across many tasks. 🧵 dpmd.ai/announcing-gem…
We’ve optimized Gemini 1.0 for three different sizes, meaning it can run on everything from data centers to mobile phones. 🔨

1️⃣ Ultra: our largest one for highly complex tasks
2️⃣ Pro: our best one for scaling across many tasks
3️⃣ Nano: our most efficient one for devices

Introducing GNoME: an AI tool that helped discover 2.2 million new crystals. 💎

Crystals are found in everything from the chips powering our phones to solar cells creating clean energy.

The model also better predicts the stability of new materials. 🧵 dpmd.ai/GNoME-AI
Graph Network for Materials Exploration (GNoME) was trained using ‘active learning’: a technique to scale up a model first trained on a small, specialized dataset. 📈

Developers can then introduce new targets, allowing machine learning to label new data with human assistance.

Meet the Google DeepMind team at #ICML2023! 👋

We’ll be presenting new research, supporting workshops with partners and hosting a range of demos.

Here’s a snapshot of the work you can hear about today at booth #109. 🧵 https://t.co/rkF1rl2vmDdpmd.ai/473bmKd
🧬 Our AI tool #AlphaFold cracked one of the biggest challenges in biology: predicting how proteins fold.

Over a million researchers are now using it to accelerate progress toward new discoveries.

Chat to the team behind the breakthrough at 10:30 HST. 📍 #ICML2023

As society becomes more digital, it’s critical to improve the code powering the world's computing.

Today in @Nature, we present AlphaDev, an AI system using reinforcement learning to discover enhanced computer science algorithms.

How does it work? 🧵 dpmd.ai/alphadev-tw 📁 AlphaDev uncovered new and improved algorithms for 𝘀𝗼𝗿𝘁𝗶𝗻𝗴, a method for ordering data.

Billions benefit from this everyday without realising it.

These underpin everything from ranking social media to how information is processed on devices. ⬇️ dpmd.ai/3OThC0G

With more powerful AI systems comes more responsibility to identify novel capabilities in models. 🔍

Our new research looks at evaluating future 𝘦𝘹𝘵𝘳𝘦𝘮𝘦 risks, which may cause harm through misuse or misalignment.

Here’s a snapshot of the work. 🧵 dpmd.ai/novel-ai-risks Under our framework, AI developers could use evaluation to uncover:

🔵 To what extent a model has certain ‘dangerous capabilities’ that could be used to threaten security, exert influence, or evade oversight
🔵 Whether it's prone to applying these to cause harm. ⬇️