Noam Brown, Sonic AI

Skip to content

Home Discover Library Ask Sonic Projects

Use with Claude or ChatGPT

Home Discover Library Ask Sonic Projects

Use with Claude or ChatGPT

Noam Brown, Sonic AI

Home/Library/Noam Brown

N

Noam Brown

Research Scientist, OpenAI · Person

199

Mentions

5

Episodes

199

Claims

Noam Brown - Research Scientist, OpenAI. Tracked across 199 mentions in podcasts and expert conversations analyzed by Sonic.

What Noam Brown has said

Noam Brown co-created an AI system, Cicero, that can strategically out-negotiate humans using natural language in the board game Diplomacy.

bullish·Lex Fridman

Noam Brown co-created the first AI system to achieve superhuman-level performance in no-limit Texas Hold'em for both heads-up and multiplayer.

bullish·Lex Fridman

Noam Brown won the World Diplomacy Championship in 2025.

neutral·Noam Brown

Noam Brown co-created an AI system, Cicero, that can strategically out-negotiate humans using natural language in the board game Diplomacy.

Expert perspectiveLex FridmanJul 28

Noam Brown co-created the first AI system to achieve superhuman-level performance in no-limit Texas Hold'em for both heads-up and multiplayer.

Expert perspectiveLex FridmanJul 28

Noam Brown won the World Diplomacy Championship in 2025.

Expert perspectiveNoam BrownJul 28

In 2022, the AI agent Cicero, developed by Noam Brown, achieved performance in the top 10% of human players in the game Diplomacy.

Expert perspectiveAlessio and SpooksJul 28

An AI trained for the 7-player version of Diplomacy using only self-play (without human data) was "destroyed" when playing against humans, even though the same training approach created a superhuman A...

Expert perspectiveNoam BrownJul 28

Modern poker AIs' value functions take player beliefs as an input, which is a different approach from how chess and Go AIs work.

Expert perspectiveNoam BrownJul 28

A researcher at the University of Toronto is working on creating AIs that can play chess in the specific style of a particular human player.

Expert perspectiveNoam BrownJul 28

The techniques developed for the Cicero AI, which rely on a well-defined value function, cannot be fully leveraged in a general chatbot setting without a similar objective function.

Expert perspectiveNoam BrownJul 28

Modern chess AIs like Stockfish play in a style that is very strong but considered "inhuman" and different from how top human players play.

Expert perspectiveNoam BrownJul 28

A top human Go player has an Elo rating of around 3,600, while the strongest version of AlphaZero is around 5,200 Elo.

Expert perspectiveNoam BrownJul 28

The counterfactual regret minimization algorithm is proven to converge to a Nash equilibrium for all games, including chess and poker.

Expert perspectiveNoam BrownJul 28

The main algorithmic improvement in Libratus (2017) over the 2015 bot was a shift from a pre-computed lookup table strategy to a real-time, search-based approach.

Expert perspectiveNoam BrownJul 28

Poker is a more complex game than Go or chess due to its imperfect information aspect.

Expert perspectiveNoam BrownJul 28

During the 2017 competition, the Libratus team provided the human players with a log of all hands played, including the bot's hidden cards, at the end of each day.

Expert perspectiveNoam BrownJul 28

Estimating an Elo-like rating in poker is significantly harder than in other games due to the game's high variance.

Expert perspectiveNoam BrownJul 28

A potential solution to the sample complexity problem in AI is to enable models to leverage general knowledge across many different domains.

Expert perspectiveNoam BrownJul 28

The version of No-Limit Texas Hold'em played by the Libratus AI has 10^161 different decision points.

Expert perspectiveNoam BrownJul 28

The models, code, and data used to train the Cicero AI for Diplomacy are being open-sourced for researchers.

Expert perspectiveNoam BrownJul 28

Daniel Negreanu constantly studies the game theory optimal (GTO) way of playing poker.

Expert perspectiveNoam BrownJul 28

The final training run for the Pluribus AI cost less than $150 on AWS.

Expert perspectiveNoam BrownJul 28

Create a free account to see Noam Brown's full intelligence report - every claim, the relationship network, and AI Q&A across all sources. No card needed.

Get started free

Back to Entities Entity Detail