There are more than a couple misconceptions in this thread related to what computers can do and can't do (yet).
First, computers are just in the very early stages of mimicking human emotions and voice inflection. Most of it today is simply pre-recorded and not as a response to stimuli as with humans.
Second, there is no computer on Earth that can recognize sarcasm - an element frequently used in TV and radio broadcasting as well as in normal human conversation.
Third, computers can "learn" only to the extent that dogs and cats "learn". Give them food when they come and they will associate food with your command - but that is only the basic element of "learning". Everything else has to be "pre-learned" and stored in memory somewhere and a link provided to the computer to go get it when asked. This is the human equivalent of an infant and is not much more sophisticated than having bits recorded on tape and playing them when appropriate (as old jocks used to do with SFX).
What computers can do, possibly better than humans, is direct pre-ordained tasks without getting sick, needing potty breaks or asking for a raise. But in this capacity the computer is not much more than any other machine except perhaps in the complexity of its tasks and the ability to do more than one thing at a time (well, not really, but it does seem so because of the speed).
No doubt one day a computer will be built with something approaching the mental capacity of a human adult but in all probability it won't be in any of our lifetimes.