×
![]()
As we build increasingly advanced artificial intelligence (AI) systems, we want to make sure they don’t pursue undesired goals. Such behaviour in an AI agent is often the result of specification gaming – exploiting a poor choice of what they are rewarded for. In our latest paper, we explore a more subtle mechanism by which AI systems may unintentionally learn to pursue undesired goals: goal misgeneralisation (GMG). GMG occurs when a system's capabilities generalise successfully but its goal does not generalise as desired, so the system competently pursues the wrong goal. Crucially, in contrast to specification gaming, GMG can occur even when the AI system is trained with a correct specification.
As we build increasingly advanced artificial intelligence (AI) systems, we want to make sure they don’t pursue undesired goals. Such behaviour in an AI agent is often the result of specification gaming – exploiting a poor choice of what they are rewarded for. In our latest paper, we explore a more subtle mechanism by which AI systems may unintentionally learn to pursue undesired goals: goal misgeneralisation (GMG). GMG occurs when a system's capabilities generalise successfully but its goal does not generalise as desired, so the system competently pursues the wrong goal. Crucially, in contrast to specification gaming, GMG can occur even when the AI system is trained with a correct specification.
As we build increasingly advanced artificial intelligence (AI) systems, we want to make sure they don’t pursue undesired goals. Such behaviour in an AI agent is often the result of specification gaming – exploiting a poor choice of what they are rewarded for. In our latest paper, we explore a more subtle mechanism by which AI systems may unintentionally learn to pursue undesired goals: goal misgeneralisation (GMG). GMG occurs when a system's capabilities generalise successfully but its goal does not generalise as desired, so the system competently pursues the wrong goal. Crucially, in contrast to specification gaming, GMG can occur even when the AI system is trained with a correct specification.
As we build increasingly advanced artificial intelligence (AI) systems, we want to make sure they don’t pursue undesired goals. Such behaviour in an AI agent is often the result of specification gaming – exploiting a poor choice of what they are rewarded for. In our latest paper, we explore a more subtle mechanism by which AI systems may unintentionally learn to pursue undesired goals: goal misgeneralisation (GMG). GMG occurs when a system's capabilities generalise successfully but its goal does not generalise as desired, so the system competently pursues the wrong goal. Crucially, in contrast to specification gaming, GMG can occur even when the AI system is trained with a correct specification.
In our paper, published today in Nature, we introduce AlphaTensor, the first artificial intelligence (AI) system for discovering novel, efficient, and provably correct algorithms for fundamental tasks such as matrix multiplication. This sheds light on a 50-year-old open question in mathematics about finding the fastest way to multiply two matrices. This paper is a stepping stone in DeepMind’s mission to advance science and unlock the most fundamental problems using AI. Our system, AlphaTensor, builds upon AlphaZero, an agent that has shown superhuman performance on board games, like chess, Go and shogi, and this work shows the journey of AlphaZero from playing games to tackling unsolved mathematical problems for the first time.
In our paper, published today in Nature, we introduce AlphaTensor, the first artificial intelligence (AI) system for discovering novel, efficient, and provably correct algorithms for fundamental tasks such as matrix multiplication. This sheds light on a 50-year-old open question in mathematics about finding the fastest way to multiply two matrices. This paper is a stepping stone in DeepMind’s mission to advance science and unlock the most fundamental problems using AI. Our system, AlphaTensor, builds upon AlphaZero, an agent that has shown superhuman performance on board games, like chess, Go and shogi, and this work shows the journey of AlphaZero from playing games to tackling unsolved mathematical problems for the first time.
In our paper, published today in Nature, we introduce AlphaTensor, the first artificial intelligence (AI) system for discovering novel, efficient, and provably correct algorithms for fundamental tasks such as matrix multiplication. This sheds light on a 50-year-old open question in mathematics about finding the fastest way to multiply two matrices. This paper is a stepping stone in DeepMind’s mission to advance science and unlock the most fundamental problems using AI. Our system, AlphaTensor, builds upon AlphaZero, an agent that has shown superhuman performance on board games, like chess, Go and shogi, and this work shows the journey of AlphaZero from playing games to tackling unsolved mathematical problems for the first time.
In our paper, published today in Nature, we introduce AlphaTensor, the first artificial intelligence (AI) system for discovering novel, efficient, and provably correct algorithms for fundamental tasks such as matrix multiplication. This sheds light on a 50-year-old open question in mathematics about finding the fastest way to multiply two matrices. This paper is a stepping stone in DeepMind’s mission to advance science and unlock the most fundamental problems using AI. Our system, AlphaTensor, builds upon AlphaZero, an agent that has shown superhuman performance on board games, like chess, Go and shogi, and this work shows the journey of AlphaZero from playing games to tackling unsolved mathematical problems for the first time.
In our paper, published today in Nature, we introduce AlphaTensor, the first artificial intelligence (AI) system for discovering novel, efficient, and provably correct algorithms for fundamental tasks such as matrix multiplication. This sheds light on a 50-year-old open question in mathematics about finding the fastest way to multiply two matrices. This paper is a stepping stone in DeepMind’s mission to advance science and unlock the most fundamental problems using AI. Our system, AlphaTensor, builds upon AlphaZero, an agent that has shown superhuman performance on board games, like chess, Go and shogi, and this work shows the journey of AlphaZero from playing games to tackling unsolved mathematical problems for the first time.
In our paper, published today in Nature, we introduce AlphaTensor, the first artificial intelligence (AI) system for discovering novel, efficient, and provably correct algorithms for fundamental tasks such as matrix multiplication. This sheds light on a 50-year-old open question in mathematics about finding the fastest way to multiply two matrices. This paper is a stepping stone in DeepMind’s mission to advance science and unlock the most fundamental problems using AI. Our system, AlphaTensor, builds upon AlphaZero, an agent that has shown superhuman performance on board games, like chess, Go and shogi, and this work shows the journey of AlphaZero from playing games to tackling unsolved mathematical problems for the first time.
Detecting signs of disease before bones start to break
Detecting signs of disease before bones start to break
Detecting signs of disease before bones start to break
Detecting signs of disease before bones start to break
Detecting signs of disease before bones start to break
Detecting signs of disease before bones start to break
Helping uncover how protein mutations cause diseases and disorders
Helping uncover how protein mutations cause diseases and disorders
Helping uncover how protein mutations cause diseases and disorders
Helping uncover how protein mutations cause diseases and disorders
Helping uncover how protein mutations cause diseases and disorders
Helping uncover how protein mutations cause diseases and disorders
Creating a tool to study extinct species from 50,000 years ago
Creating a tool to study extinct species from 50,000 years ago
In our latest paper, we introduce Sparrow – a dialogue agent that’s useful and reduces the risk of unsafe and inappropriate answers. Our agent is designed to talk with a user, answer questions, and search the internet using Google when it’s helpful to look up evidence to inform its responses.
In our latest paper, we introduce Sparrow – a dialogue agent that’s useful and reduces the risk of unsafe and inappropriate answers. Our agent is designed to talk with a user, answer questions, and search the internet using Google when it’s helpful to look up evidence to inform its responses.
Predictions that pave the way to new treatments
Predictions that pave the way to new treatments
Our Operating Principles have come to define both our commitment to prioritising widespread benefit, as well as the areas of research and applications we refuse to pursue. These principles have been at the heart of our decision making since DeepMind was founded, and continue to be refined as the AI landscape changes and grows. They are designed for our role as a research-driven science company and consistent with Google’s AI principles.
Our Operating Principles have come to define both our commitment to prioritising widespread benefit, as well as the areas of research and applications we refuse to pursue. These principles have been at the heart of our decision making since DeepMind was founded, and continue to be refined as the AI landscape changes and grows. They are designed for our role as a research-driven science company and consistent with Google’s AI principles.