The classic proposal for deciding whether a computer program has attained human-like intelligence is the Turing test: can it maintain a conversation (by text, not necessarily speech) with a real human being after which the human is unable to tell that he or she has been conversing with a program?
So far, no computer program has come close, but there are people who like to write programs that carry on conversations. These are now called chatterbots or chatbots. The fourth annual Chatterbox Challenge is now underway. You can engage any of the contestants in conversation and participate in the voting. Regrettably, the contest is limited to anglophone bots and humans.