How It Works
Markov Chain
A model of a sequence in which the next step depends only on the current state, not on the whole history that led there. Fed text, it produces sentences that are locally plausible and globally meaningless — the ancestor of today’s language models, and a useful reminder of what “predict the next bit” gets you on its own.
Origin
Introduced by Russian mathematician Andrey Markov in 1906, who later demonstrated it on the letters of Pushkin’s “Eugene Onegin.” Text generators built on it long predate AI chatbots — the Mark V. Shaney bot posted Markov-generated Usenet messages in the 1980s.