People talking about "LLMs can't make new information" is so fucking funny. Ah yeah, AlphaZero hasn't found new chess strategies. Protein folding models don't make information.
December 21, 2025 at 4:20 PM
People talking about "LLMs can't make new information" is so fucking funny. Ah yeah, AlphaZero hasn't found new chess strategies. Protein folding models don't make information.
alphazero doesnt want anything
December 26, 2024 at 7:26 PM
alphazero doesnt want anything
alphazero smth smth... no pun intended...
September 22, 2026 at 7:13 PM
alphazero smth smth... no pun intended...
Deep learning ≠ autoregressive language models en.m.wikipedia.org/wiki/AlphaZero
AlphaZero - Wikipedia
en.m.wikipedia.org
July 26, 2025 at 1:26 AM
Deep learning ≠ autoregressive language models en.m.wikipedia.org/wiki/AlphaZero
Alphazero go bots are superhuman, but the adversarial attacks reveal that their “understanding“ of the game is alien and imperfect. Does the same thing happen when the base model is an LLM?
July 4, 2026 at 8:23 PM
Alphazero go bots are superhuman, but the adversarial attacks reveal that their “understanding“ of the game is alien and imperfect. Does the same thing happen when the base model is an LLM?
I'm delighted to announce the release of OpenSpiel 2.0! ♟️🎲♦️🎉
Structured types for states, observations, and actions, standard trajectories (based on JSON), 19 new games, AlphaZero ported to JAX, Windows PyPI support, language model fine-tuning examples and an MCP server (demo below 🤩👇)!
🧵 1/N
Structured types for states, observations, and actions, standard trajectories (based on JSON), 19 new games, AlphaZero ported to JAX, Windows PyPI support, language model fine-tuning examples and an MCP server (demo below 🤩👇)!
🧵 1/N
July 14, 2026 at 4:06 PM
I'm delighted to announce the release of OpenSpiel 2.0! ♟️🎲♦️🎉
Structured types for states, observations, and actions, standard trajectories (based on JSON), 19 new games, AlphaZero ported to JAX, Windows PyPI support, language model fine-tuning examples and an MCP server (demo below 🤩👇)!
🧵 1/N
Structured types for states, observations, and actions, standard trajectories (based on JSON), 19 new games, AlphaZero ported to JAX, Windows PyPI support, language model fine-tuning examples and an MCP server (demo below 🤩👇)!
🧵 1/N
In contrast, the true “AlphaZero moment” arrived in 2018 with BERT, which redefined NLP by enabling models to better understand context, simplifying development pipelines, and reducing dependence on extensive labeled datasets.
August 22, 2025 at 4:09 AM
In contrast, the true “AlphaZero moment” arrived in 2018 with BERT, which redefined NLP by enabling models to better understand context, simplifying development pipelines, and reducing dependence on extensive labeled datasets.
AlphaZero: Distributed Reinforcement Learning with Monte Carlo Search Tree! #BigData #Analytics #AI #MachineLearning #DataScience #IoT #IIoT #Python #RStats #TensorFlow #JavaScript #ReactJS #CloudComputing #Serverless #DataScientist #Linux #Programming #Coding #100DaysofCode
geni.us/AlphaZero
geni.us/AlphaZero
September 16, 2026 at 5:26 AM
AlphaZero: Distributed Reinforcement Learning with Monte Carlo Search Tree! #BigData #Analytics #AI #MachineLearning #DataScience #IoT #IIoT #Python #RStats #TensorFlow #JavaScript #ReactJS #CloudComputing #Serverless #DataScientist #Linux #Programming #Coding #100DaysofCode
geni.us/AlphaZero
geni.us/AlphaZero
On the one hand, the meaning of ‘AI’ has largely been pegged to concrete LLM based systems offered by the big players, erasing the wider range of techniques beyond deep neural nets and even DNNs beyond the transformer/diffusion paradigm (e.g., Alphazero, Alphafold, etc.).
May 19, 2026 at 7:54 AM
On the one hand, the meaning of ‘AI’ has largely been pegged to concrete LLM based systems offered by the big players, erasing the wider range of techniques beyond deep neural nets and even DNNs beyond the transformer/diffusion paradigm (e.g., Alphazero, Alphafold, etc.).
im just confused.. google already owns the strongest chess engine in existence. why not just have gemini ask alphazero? can it not even do that??
July 14, 2025 at 5:27 PM
im just confused.. google already owns the strongest chess engine in existence. why not just have gemini ask alphazero? can it not even do that??
Mer om AlphaZero, föregångaren i att 'träna' en AI i att spela spel med enkla regler: en.wikipedia.org/wiki/AlphaZero
AlphaZero - Wikipedia
en.wikipedia.org
October 22, 2025 at 3:00 PM
Mer om AlphaZero, föregångaren i att 'träna' en AI i att spela spel med enkla regler: en.wikipedia.org/wiki/AlphaZero
One of my grand interpretability goals is to improve human scientific understanding by analyzing scientific discovery models, but this is the most convincing case yet that we CAN learn from model interpretation: Chess grandmasters learned new play concepts from AlphaZero's internal representations.
Bridging the Human-AI Knowledge Gap: Concept Discovery and Transfer in AlphaZero
Artificial Intelligence (AI) systems have made remarkable progress, attaining super-human performance across various domains. This presents us with an opportunity to further human knowledge and improv...
arxiv.org
January 27, 2025 at 9:43 PM
One of my grand interpretability goals is to improve human scientific understanding by analyzing scientific discovery models, but this is the most convincing case yet that we CAN learn from model interpretation: Chess grandmasters learned new play concepts from AlphaZero's internal representations.
Der AlphaZero-Moment für Wetterprognosen
A new computer model to predict the weather built by Google and powered by AI consistently outperforms and is many times faster than government models that have existed for decades and involved hundreds of millions of dollars in investment, according to a study published Tuesday.
Why your weather forecasts may soon become more accurate
Google has produced a weather forecasting model using artificial intelligence with better accuracy, faster speed and lower costs, study finds.
www.washingtonpost.com
November 14, 2023 at 10:04 PM
Der AlphaZero-Moment für Wetterprognosen
Many compare the 2017 paper “Attention Is All You Need” to an “AlphaZero moment” in AI - but that comparison is somewhat misleading. While the Transformer architecture it introduced is now foundational, it initially drew limited attention within the NLP community.
August 22, 2025 at 4:09 AM
Many compare the 2017 paper “Attention Is All You Need” to an “AlphaZero moment” in AI - but that comparison is somewhat misleading. While the Transformer architecture it introduced is now foundational, it initially drew limited attention within the NLP community.
*There hasn't yet been an "AlphaZero" AI that's entirely divorced from any human databases and that self-teaches about the nature of the universe
*Likely it would need to be a robot of some kind, embodied, lots of sensors and manipulators
*Maybe thousands of them, a cognitive anthill
*Likely it would need to be a robot of some kind, embodied, lots of sensors and manipulators
*Maybe thousands of them, a cognitive anthill
February 26, 2026 at 7:09 AM
*There hasn't yet been an "AlphaZero" AI that's entirely divorced from any human databases and that self-teaches about the nature of the universe
*Likely it would need to be a robot of some kind, embodied, lots of sensors and manipulators
*Maybe thousands of them, a cognitive anthill
*Likely it would need to be a robot of some kind, embodied, lots of sensors and manipulators
*Maybe thousands of them, a cognitive anthill
My brilliant daughter @lisaschut.bsky.social joined 🦋 today!
She is first author of the Google DeepMind paper that shows that AlphaZero plays chess using unique superhuman concepts and improved solving performance of Gukesh, Kramnik and Hou Yifan by teaching these concepts
arxiv.org/abs/2310.16410
Bridging the Human-AI Knowledge Gap: Concept Discovery and Transfer in AlphaZero
Artificial Intelligence (AI) systems have made remarkable progress, attaining super-human performance across various domains. This presents us with an opportunity to further human knowledge and improv...
arxiv.org
November 16, 2024 at 12:31 AM
My brilliant daughter @lisaschut.bsky.social joined 🦋 today!
She is first author of the Google DeepMind paper that shows that AlphaZero plays chess using unique superhuman concepts and improved solving performance of Gukesh, Kramnik and Hou Yifan by teaching these concepts
arxiv.org/abs/2310.16410
AlphaZero, Google DeepMind's breakthrough algorithm, mastered chess and Go without human input.🧠
Join us in Sydney, where you'll use Goose to build the AlphaZero algorithm and adapt it to a domain of your choice.
Spots are limited. Sign up 👇
lu.ma/bb6xh9zs
Join us in Sydney, where you'll use Goose to build the AlphaZero algorithm and adapt it to a domain of your choice.
Spots are limited. Sign up 👇
lu.ma/bb6xh9zs
Building AlphaZero with Project Goose · Luma
For anyone wanting to take their AI skills to the superhuman level. This workshop is for you.
Note: this workshop repeats for 4 weeks in a row!
In 2017 Google…
lu.ma
July 11, 2025 at 2:00 AM
AlphaZero, Google DeepMind's breakthrough algorithm, mastered chess and Go without human input.🧠
Join us in Sydney, where you'll use Goose to build the AlphaZero algorithm and adapt it to a domain of your choice.
Spots are limited. Sign up 👇
lu.ma/bb6xh9zs
Join us in Sydney, where you'll use Goose to build the AlphaZero algorithm and adapt it to a domain of your choice.
Spots are limited. Sign up 👇
lu.ma/bb6xh9zs
IMO it is well known, that MCTS was around well before AlphaZero.
December 20, 2024 at 2:04 PM
IMO it is well known, that MCTS was around well before AlphaZero.
But if you are curious about what we do know about such systems, this paper is awesome arxiv.org/abs/2310.16410
Bridging the Human-AI Knowledge Gap: Concept Discovery and Transfer in AlphaZero
Artificial Intelligence (AI) systems have made remarkable progress, attaining super-human performance across various domains. This presents us with an opportunity to further human knowledge and improv...
arxiv.org
August 10, 2025 at 2:39 AM
But if you are curious about what we do know about such systems, this paper is awesome arxiv.org/abs/2310.16410
I had someone explaining to me that "we understood" AlphaZero because we know what algorithm was used to train it
August 13, 2025 at 10:02 PM
I had someone explaining to me that "we understood" AlphaZero because we know what algorithm was used to train it
Después de leer 'MANIAC', de Benjamín Labatut, he de decir que la única IA que me aterra se llama AlphaZero.
May 15, 2025 at 9:18 AM
Después de leer 'MANIAC', de Benjamín Labatut, he de decir que la única IA que me aterra se llama AlphaZero.
Check out this interview with @nenadtomasev.bsky.social about AlphaZero at the Chess World Championship!
www.youtube.com/watch?v=ppiW...
www.youtube.com/watch?v=ppiW...
Game 2 Commentary with GM David Howell and IM Jovanka Houska | FIDE World Championship Match 2024
YouTube video by FIDE chess
www.youtube.com
November 26, 2024 at 1:44 PM
Check out this interview with @nenadtomasev.bsky.social about AlphaZero at the Chess World Championship!
www.youtube.com/watch?v=ppiW...
www.youtube.com/watch?v=ppiW...
自作ボードゲームの対戦相手、チャッピーだと多分怪しいからAlphaZeroとかいじってみるかな…🤔
October 7, 2025 at 12:36 PM
自作ボードゲームの対戦相手、チャッピーだと多分怪しいからAlphaZeroとかいじってみるかな…🤔
I made a 3 part YouTube series on *AlphaZero Explained* which explains the math/algorithm behind Google DeepMind's AI that achieved superhuman performance in Chess & Go.
All 3 videos come with an *interactive online Python notebook* using Connect 4 as an example.
PART 1: youtu.be/mYZlV44UzLc
All 3 videos come with an *interactive online Python notebook* using Connect 4 as an example.
PART 1: youtu.be/mYZlV44UzLc
AlphaZero Explained 1: How it Solves "Explore vs Exploit" [UCB Algorithm]
YouTube video by Prof Mihai Nica
youtu.be
July 2, 2026 at 6:11 PM
I made a 3 part YouTube series on *AlphaZero Explained* which explains the math/algorithm behind Google DeepMind's AI that achieved superhuman performance in Chess & Go.
All 3 videos come with an *interactive online Python notebook* using Connect 4 as an example.
PART 1: youtu.be/mYZlV44UzLc
All 3 videos come with an *interactive online Python notebook* using Connect 4 as an example.
PART 1: youtu.be/mYZlV44UzLc