One of us, let us call him B, takes up contact with a machine A (by means of a data processing or teletype medium). C a censor is there to supervise the dialogue. A simulates a human being, and so does B; and now C must decide which is the human being and which the machine. P and L, profit and loss (the market), are also introduced into the equation.
What of a pulsating machine A that whenever it betrays itself (either by making or, on the contrary, not making a mistake), it improves its performance? It learns and learns. At what point will C be unable to decide between A nd B, who is the human being and who the pulsating machine?