Demis talks about AlphaGo choosing the optimal on a board with 10^170 positions but we already know that AlphaGo was very suboptimal. Which raises the question of how we might ever know how good AlphaGo is in any absolute sense.
November 22, 2024 at 11:10 PM
Demis talks about AlphaGo choosing the optimal on a board with 10^170 positions but we already know that AlphaGo was very suboptimal. Which raises the question of how we might ever know how good AlphaGo is in any absolute sense.
if we have a chatgpt/alphago release moment for robotics it will be someone figuring out what training objective you can actually scale for moving around in space
October 10, 2025 at 8:29 PM
if we have a chatgpt/alphago release moment for robotics it will be someone figuring out what training objective you can actually scale for moving around in space
A better analogy would be AlphaGo in 2016, during Move 37.
It used Monte Carlo Tree search, which is a super cool subfield of AI research.
It used Monte Carlo Tree search, which is a super cool subfield of AI research.
September 17, 2026 at 12:28 AM
A better analogy would be AlphaGo in 2016, during Move 37.
It used Monte Carlo Tree search, which is a super cool subfield of AI research.
It used Monte Carlo Tree search, which is a super cool subfield of AI research.
AlphaGo - The Movie | Full award-winning documentary
YouTube video by Google DeepMind
m.youtube.com
December 16, 2024 at 3:39 AM
the illustrative example here isn't really chatgpt, it's alphago
May 24, 2025 at 5:33 PM
the illustrative example here isn't really chatgpt, it's alphago
We evidently don't feel that AlphaGo Zero is an "individual" or "acted" on the world simply because it developed novel (and effective!) strategies. We presumably would still not feel this way if we hooked AlphaGo up to a system that led its moves to have real world effects. Is this right?
September 18, 2026 at 1:58 PM
We evidently don't feel that AlphaGo Zero is an "individual" or "acted" on the world simply because it developed novel (and effective!) strategies. We presumably would still not feel this way if we hooked AlphaGo up to a system that led its moves to have real world effects. Is this right?
How to quantify the impact of AI on long-run cultural evolution? Published today, I give it a go!
400+ years of strategic dynamics in the game of Go (Baduk/Weiqi), from feudalism to AlphaGo!
400+ years of strategic dynamics in the game of Go (Baduk/Weiqi), from feudalism to AlphaGo!
September 16, 2025 at 2:04 PM
How to quantify the impact of AI on long-run cultural evolution? Published today, I give it a go!
400+ years of strategic dynamics in the game of Go (Baduk/Weiqi), from feudalism to AlphaGo!
400+ years of strategic dynamics in the game of Go (Baduk/Weiqi), from feudalism to AlphaGo!
Staying in Bisbee is like stumbling into a novel. While the bar jam band took a break, the flute player ordered a drink and chatted us up... turns out he's a retired math professor whose son (at the time a high-level Go player) acquired a form of AI psychosis when AlphaGo became unbeatable
May 7, 2026 at 2:59 PM
Staying in Bisbee is like stumbling into a novel. While the bar jam band took a break, the flute player ordered a drink and chatted us up... turns out he's a retired math professor whose son (at the time a high-level Go player) acquired a form of AI psychosis when AlphaGo became unbeatable
[apropos of none of this] alphago was TEN YEARS AGO? what the fuck
March 5, 2026 at 3:34 AM
[apropos of none of this] alphago was TEN YEARS AGO? what the fuck
1) What do you like about AI and why?
most fundamentally I find the latent space incredibly fascinating to explore. the tech has intrigued me profoundly since it was weird dogslug shoggoths from DeepDream; I think AlphaGo taught us something about the game of Go we'd have not learned otherwise
most fundamentally I find the latent space incredibly fascinating to explore. the tech has intrigued me profoundly since it was weird dogslug shoggoths from DeepDream; I think AlphaGo taught us something about the game of Go we'd have not learned otherwise
September 24, 2026 at 9:52 PM
1) What do you like about AI and why?
most fundamentally I find the latent space incredibly fascinating to explore. the tech has intrigued me profoundly since it was weird dogslug shoggoths from DeepDream; I think AlphaGo taught us something about the game of Go we'd have not learned otherwise
most fundamentally I find the latent space incredibly fascinating to explore. the tech has intrigued me profoundly since it was weird dogslug shoggoths from DeepDream; I think AlphaGo taught us something about the game of Go we'd have not learned otherwise
Podcast CB SyR 550: AlphaGo de Deepmind, momificación egipcia y teoría de la consciencia francis.naukas.com/2026/03/20/p...
Podcast CB SyR 550: AlphaGo de Deepmind, momificación egipcia y teoría de la consciencia - La Ciencia de la Mula Francis
Te recomiendo disfrutar del episodio 550 del podcast Coffee Break: Señal y Ruido [iVoox A, iVoox B; ApplePod A, ApplePod B], titulado “Entrevista DeepMind; AlphaGo y AlphaFold; Egipto; Consciencia”, 1...
francis.naukas.com
March 20, 2026 at 9:17 PM
Podcast CB SyR 550: AlphaGo de Deepmind, momificación egipcia y teoría de la consciencia francis.naukas.com/2026/03/20/p...
David Silver has a new billion-dollar company that aims to build AI “superlearners.”
The Man Behind AlphaGo Thinks AI Is Taking the Wrong Path
David Silver has a new billion-dollar company that aims to build AI “superlearners.”
wrd.cm
May 2, 2026 at 2:45 PM
David Silver has a new billion-dollar company that aims to build AI “superlearners.”
Thore Graepel, an AlphaGo co-creator who left Google DeepMind this summer, is seeking tens of millions of dollars for a new AI reasoning startup
Ex-DeepMind Researcher Who Worked On AlphaGo Seeks Funds for Startup Working on AI Reasoning
A co-creator of AlphaGo, the landmark AI program from Google DeepMind, is the latest prominent researcher from the company to depart and seek significant funding for a new startup, according to people...
www.bloomberg.com
September 24, 2026 at 11:06 PM
Thore Graepel, an AlphaGo co-creator who left Google DeepMind this summer, is seeking tens of millions of dollars for a new AI reasoning startup
alphago jailbreak!
Fascinating: humans unexpectedly regained their advantage in Go because they created situations AIs never had seen in their training data https://franklantz.substack.com/p/the-afterlife-of-go?utm_source=post-email-title&publication_id=1435607&post_id=117118804&isFreemail=true&utm_medium=email
May 15, 2023 at 6:22 PM
alphago jailbreak!
How AlphaGo like architectures can explain human insight. Out now in Cognition!
my paper with max, @maxkw.bsky.social, tuomas, and @fierycushman.bsky.social out in cognition at long last www.sciencedirect.com/science/arti...
We explain why humans and successful AI planners both fail on a certain kind of problem that we might describe as requiring insight or creativity
We explain why humans and successful AI planners both fail on a certain kind of problem that we might describe as requiring insight or creativity
Similar failures of consideration arise in human and machine planning
Humans are remarkably efficient at decision making, even in “open-ended” problems where the set of possible actions is too large for exhaustive evalua…
www.sciencedirect.com
March 14, 2025 at 3:26 PM
How AlphaGo like architectures can explain human insight. Out now in Cognition!
something something if ur playing a game of go ur prolly in an alphago sim something
July 16, 2026 at 12:33 AM
something something if ur playing a game of go ur prolly in an alphago sim something
So *I'm* not saying AlphaGo or an LLM/agent is an individual, but more so trying to understand why we clearly don't think of AlphaGo the way we think of an LLM in relation to its environment.
September 18, 2026 at 2:12 PM
So *I'm* not saying AlphaGo or an LLM/agent is an individual, but more so trying to understand why we clearly don't think of AlphaGo the way we think of an LLM in relation to its environment.
*I'm looking for examples for radically unexpected AI "chain-of-thought" chains of action. Like AlphaGo's famous surprising move against Lee Sedol on the go-board, but for kinetic activities in the real world.
en.wikipedia.org/wiki/AlphaGo...
en.wikipedia.org/wiki/AlphaGo...
AlphaGo versus Lee Sedol - Wikipedia
en.wikipedia.org
January 19, 2026 at 7:52 AM
*I'm looking for examples for radically unexpected AI "chain-of-thought" chains of action. Like AlphaGo's famous surprising move against Lee Sedol on the go-board, but for kinetic activities in the real world.
en.wikipedia.org/wiki/AlphaGo...
en.wikipedia.org/wiki/AlphaGo...
"alphago and anything resembling it is not safe to give real power to" is true but this is not a property of things which can be made, it is a property of the alphago training regime
September 10, 2026 at 12:06 AM
"alphago and anything resembling it is not safe to give real power to" is true but this is not a property of things which can be made, it is a property of the alphago training regime
Have felt for a long time that the 1 loss of AlphaGo to Lee Sedol was at least as informative as its 4 wins. Limitations are just more interesting than capabilities, once you've got them.
July 13, 2026 at 12:18 AM
Have felt for a long time that the 1 loss of AlphaGo to Lee Sedol was at least as informative as its 4 wins. Limitations are just more interesting than capabilities, once you've got them.
GPT-3 came out closer to AlphaGo than to today.
September 12, 2024 at 10:31 AM
GPT-3 came out closer to AlphaGo than to today.
Hey mathematicians:
Separate from whether you think it's possible or likely, how do you think you would be psychologically affected if there was a superhuman machine for both proposing and solving theorems? Meaning basically that, vis-a-vis mathematics, you were like a Go player after AlphaGo.
Separate from whether you think it's possible or likely, how do you think you would be psychologically affected if there was a superhuman machine for both proposing and solving theorems? Meaning basically that, vis-a-vis mathematics, you were like a Go player after AlphaGo.
July 18, 2025 at 5:55 PM
Hey mathematicians:
Separate from whether you think it's possible or likely, how do you think you would be psychologically affected if there was a superhuman machine for both proposing and solving theorems? Meaning basically that, vis-a-vis mathematics, you were like a Go player after AlphaGo.
Separate from whether you think it's possible or likely, how do you think you would be psychologically affected if there was a superhuman machine for both proposing and solving theorems? Meaning basically that, vis-a-vis mathematics, you were like a Go player after AlphaGo.
We don’t understand systems we build. Even simple systems. We don’t even really understand how a tiny neural net like AlphaGo is so good at Go!
August 10, 2025 at 2:39 AM
We don’t understand systems we build. Even simple systems. We don’t even really understand how a tiny neural net like AlphaGo is so good at Go!
AlphaGo can't be beaten at Go. Imagine its skill and efficiency at deciding who lives and who dies. The Torment Nexus is here.
This orangutan can't be beaten at climbing up a tree. Imagine its skill and poise as our next secretary of education.
November 24, 2025 at 12:24 PM
AlphaGo can't be beaten at Go. Imagine its skill and efficiency at deciding who lives and who dies. The Torment Nexus is here.
Woah that's not a good description of AlphaGo
June 20, 2025 at 11:45 PM
Woah that's not a good description of AlphaGo