Eleusis is an induction card game that mirrors the scientific process. Is there an AI that can play Eleusis?

From a quick search this paper suggests the answer is 'no', but the paper dates from 1980, and I am not sure if newer developments have happened.

  • I believe whether an AI can play this game depends on whether there exists some upper bound for what the rules can involve (cards must touch the previous card, players must place with their left hand, cards can only be played on Tuesdays, etc.) Without that, I don't think it's possible to give an AI 'all' the data it needs. Commented Feb 22 at 3:46

2 Answers 2


It's easy to write an AI to play Eleusis, as long as you don't need it to play well

As a player: on the AI's turn, randomly pick a card in their hand and try and play it.

As a judge: choose the rule "the card played must be exactly N higher or lower than the previous card" where N is chosen randomly from 1 to 6 when the rule is picked

Will this AI win? Probably not, unless the human players are really bad. But it's functional. It's fairly simple to devise a more successful AI than this as well. Can an AI outperform a human? Well, that depends on how good the human is, doesn't it? I bet I can write an AI that outperforms my cousin. My cousin is 4, but my AI is younger than that!

  • "Can an AI outperform a human? Well, that depends on how good the human is, doesn't it?" not really, state of the art chess AI can outperform human regardless of how good the human is. Commented Feb 28 at 9:03

According to this paper, "several algorithms have been developed to play Eleusis":

M. J. Berry. APL and the search for truth: A set of functions to play NewEleusis. In APL ’81: Proceedings of the international conference on APL, pages47–53, New York, NY, USA, 1981. ACM

T. G. Diettrich and R. S. Michalski. Learning to predict sequences. InMachine Learning: An Artificial Intelligence Approach, volume II. MorganKaufmann, 1989

R. S. Michalski, H. Ko, and K. Chen. SPARC/E(V.2): A eleusis rule generatorand player. Report ISG85-11, UIUCDC-F-85941, Department of ComputerScience, University of Illinois, 1985

You must log in to answer this question.