Loading...

AlphaZero vs. Stockfish

Google-funded DeepMind has now shown that their AlphaZero algorithm can't just be used to beat Go but can outplay the best existing chess engines at chess and shogi. Their algorithm took a mere 4 hours of playing games against itself to teach itself to play chess at a level superior to Stockfish 8! In 100 games AlphaZero scored 25 wins and 25 draws with White, while with Black it scored 3 wins and 47 draws. It didn't lose a game, with the final score 64:36. Here you can replay 10 example games by our new computer overlord. Read the DeepMind paper (pdf): arxiv.org/pdf/1712.01815.pdf
Register
or

Create your free account now to get started!

By clicking ‘Register’ you agree to our terms and conditions and confirm you have read our privacy policy, including the section on the use of cookies.

Lost your password? We'll send you a link to reset it!

After submitting this form you'll receive an email with the reset password link. If you still can't access your account please contact our customer service.