Mastermind and an AI that stomps you pretty hard unless you're very careful.
There’s a sincere pleasure to insight, even if it isn’t yours. The sudden shock of knowledge makes me laugh, which I think is normal, despite the occasional social indelicacy of it.
I’ve been noodling making something where a computer learns how to play a game. Ideally through some inscrutably complex, semi-random method, a la MarI/O. I don’t have the vast data set or unique machine learning insight to prepare an interesting (if less effective) knock-off of AlphaGo, even for a simpler game. And marveling at the inscrutability of a superior machine player is almost as fun as marveling at the mind of a superior human player, which used to be reasonably common.
Isn’t it weird that I can recall more than one instance of people describing how legendary human chess players think, but I’ve read much more about DeepBlue and AlphaGo and all I know for sure is that they are better than us? I guess they’d be seen as a lateral thinker – which is to say, it looks for opportunities in places even experts normally don’t – which is massively unsurprising and nearly a tautology in these circumstances.
Anyhow, if you want to peer into the greatness of a computer mind, feel free to play my version of the classic board game MasterMind -- and then compare the computer’s ability to solve the same code (you can also play reverse MasterMind, where you choose the code, a slightly less engaging role in the classical board game). With practice, you’ll only get slightly but consistently outwitted. What’s the key behind the computer’s genius? It makes a simple first move Computer Science Legend Donald Knuth says is required for ideal play, but after that it just chooses a random plausible solution.
And it destroys me, consistently (on best of three – I have beaten it a couple times).
Makes me laugh.