A360
Key Takeaways✦ Atlas AI
01
AI self-play fails impartial games.
02
Nim game complexity halts AI learning.
03
New AI training methods are needed.
Atlas AI
A recent study published in Machine Learning by Bei Zhou and Soren Riis indicates that the self-play training methods used for advanced AI systems like Google DeepMind's AlphaGo and AlphaZero are ineffective for a category of games known as impartial games. This limitation was demonstrated using the game of Nim, where AI performance significantly degraded as game complexity increased.
For a five-row Nim board, the AI showed improvement over 500 training iterations, but for a seven-row board, performance gains ceased after 500 iterations, becoming indistinguishable from random move selection.
Impartial games, unlike games such as chess, feature shared pieces and identical rules for both players. Nim is a critical example, as any position in an impartial game can be represented by a Nim configuration. Optimal moves in impartial games can be determined by a parity function, which indicates the winning player.
The AlphaGo approach, which relies on associating board configurations with win probabilities through self-play and random sampling, struggles to develop this functional understanding.
The research highlights a fundamental failure mode in current AI training paradigms when applied to environments where optimal strategies are derived from underlying mathematical functions rather than complex, probabilistic evaluations. This suggests a need for alternative training methodologies to address these specific types of game structures, impacting the broader application of AI in problem-solving.
Related Articles
About this story
Atlas360 covers Technology as part of a broader effort to give international readers fast, source-checked context on global affairs. Our newsroom monitors original reporting from wire services, accredited correspondents and verified eyewitness accounts, then re-summarises the most important facts in clear, plain-language English so that you can understand both what happened and why it matters.
Every published article on Atlas360 is reviewed for accuracy, balance and timeliness before it reaches the homepage. When new information emerges — for example a correction from an official source, a casualty update, or a clarifying statement from a named spokesperson — we update the story in place and keep the original publication time so readers can track how a developing situation evolves.
If you want to keep following Technology, you can browse the related coverage at the foot of this page, subscribe to the Atlas360 newsletter for a daily roundup, or open the relevant topic page where every story we have published on the subject is listed in reverse chronological order. Reader signals from the community feed also shape which threads we keep reporting on.