The #AlphaZero paper was cool for many reasons, but one reason that doesn't get talked about enough is the image attached to this post.

It's pretty cool to see how the engine went through phases of preferring different openings.
One interesting one was my weapon of choice, the Caro-Kann. For quite a while it seemed to like it a lot, then near the end it simply stopped playing it. I kind of wonder if it found some sort of refutation.

#Chess

If #AI start writing undetectable malware (https://www.techspot.com/news/98202-researcher-used-chatgpt-create-dangerous-data-stealing-malware.html) then what's the long term solution? Right now we probably don't have adequate defenses against many different vectors, because we haven't imagined them yet. But advanced AI might, just like how #AlphaZero discovered new strategies for Chess and Go.

Maybe it's a time for #OpenSource to shine, coupled with firewall AI that reads and analyzes the code before running.

#security #programming

DeepMind has done really well at training game-playing neural network systems such as #AlphaZero. Who is exploring combining the ideas of AlphaZero's self-play training with LLM networks to help #chatgpt avoid hallucinations?

#FreeCodeCamp and this person rock:

The video course teaches how to code an AlphaZero algorithm from scratch to play Tic Tac Toe and Connect Four. The course is divided into ten sections, starting with an introduction to the course and an overview of the #AlphaZero algorithm.
#machinelearning

https://www.freecodecamp.org/news/code-alphazero-machine-learning-algorithm/

DeepMind has open-sourced the core of #AlphaGo and #AlphaZero. It’s a library called mccx and provides JAX-native implementation of Monte Carlo Tree Search

https://github.com/deepmind/mctx

I wrote this four years ago. Today, we're much closer to the reality I described than I thought we'd be…

#AlphaZero, #machine #learning, and the #future of #work https://is.gd/10zWff

Acquisition of chess knowledge in AlphaZero
https://www.pnas.org/doi/10.1073/pnas.2206625119

We analyze the knowledge acquired by AlphaZero, a neural network engine that learns chess solely by playing against itself yet becomes capable of outperforming human chess players.

#ai #neural_network #alphazero #chess #behavioral_analysis

I feel like I just beat #AlphaZero at chess and it kicked over the chessboard.

Not only did #chatgpt have to apologize, it tried gaslighting its poor judgement about history and then knocked over the table....

So while I was driving to Melbourne and back a couple of weeks ago I was listening to a biography of Alan #Turing.
More recently while I have been laid up with #covid I have been watching GothamChess' analysis of #chess games on his YouTube channel.
Turing was always interested in chess playing and with a friend discovered the min-max strategy for selecting chess moves. However Turing would have exactly understood how the latest generation of chess playing programs such as #AlphaZero and Leela operate. When he was at Hut 8 in Bletchly, Turing set up a procedure using Bayesian methods to evaluate potential cribs to use with the Bombes. These had their own unit of measure the deciban. Despite being mechanically different and cloaked in modern machine learning terms evaluating a move against the probability of eventually winning the game is what Alpha Zero and Leela do. I think he would have liked that.

So while I was driving to Melbourne and back a couple of weeks ago I was listening to a biography of Alan #Turing.
More recently while I have been laid up with #covid I have been watching GothamChess' analysis of #chess games on his YouTube channel.
Turing was always interested in chess playing and with a friend discovered the min-max strategy for selecting chess moves. However Turing would have exactly understood how the latest generation of chess playing programs such as #AlphaZero and Leela operate. When he was at Hut 8 in Bletchly, Turing set up a procedure using Bayesian methods to evaluate potential cribs to use with the Bombes. These had their own unit of measure the deciban. Despite being mechanically different and cloaked in modern machine learning terms evaluating a move against the probability of eventually winning the game is what Alpha Zero and Leela do. I think he would have liked that.

There have been a few key moments in the recent history of AI: the Hinton et al's "vision" neural network; Watson's Jeopardy! win; AlphaZero's chess prowess. Any others you would add? Is OpenAI's chatbot in the same category? #ai #chatbot #alphazero #watson

RT @demishassabis@twitter.com

It was a real honour to closely collaborate with ex-World Chess Champion Vladimir Kramnik on this work, which I think is one of the most thorough analyses yet of the representations and knowledge a complex neural network system like #AlphaZero builds up. Published in PNAS today. https://twitter.com/DeepMind/status/1593213217081528322

🐦🔗: https://twitter.com/demishassabis/status/1593356855719477249

@gmmds I recently saw a technical article from the folks at #AlphaZero (more terms of art than a museum). Chess playing AI (a term I use loosely) has always seemed to me like using a group of GMs to collude to play each game. #chess #software

#DeepMind’s game-playing #AI has beaten a 50-year-old record in #ComputerScience
The new version of #AlphaZero discovered a faster way to do matrix multiplication, a core problem in computing that affects thousands of everyday computer tasks.
https://bit.ly/3C7tYu2

I have now been following both #Leela and #AlphaZero chess AIs for years, and what is most fascinating is that they do NOT play as automatons, but, quite the contrary, take beautiful, inspired risks which will not always pan out, but produce magical games.

Perhaps art is nothing else than superior thought? And where does that leave self-awareness and a (silly) #TuringTest?

Published games:
https://www.youtube.com/playlist?list=PLQsLDm9Rq9bEvlHwbsWlt5sta3M2t0aGe

https://www.deepmind.com/
https://lczero.org/ #mastodontwittercrosspost #AI

#AlphaZero #AI beats champion #chess #program after teaching itself in four hours | #Technology | The Guardian

https://www.theguardian.com/technology/2017/dec/07/alphazero-google-deepmind-ai-beats-champion-program-teaching-itself-to-play-four-hours

#ComputerLearning