What would need a computer to posses a Turning test? or Define natural language processing & automated reasoning
||Artificial Intelligence and Neural Network
A computer needn’t have context awareness, sophisticated language understanding, nor any other features that resemble “human” per se. It must only be able to present textual responses to questions which cannot be distinguished from those which a human might write. That’s an easier problem to solve than building an artificial mind.
A computer is a very different tool than a brain. With different tools, different approaches are often needed to yield optimal results. If you told me to remove a screw with a hammer, would I try to unscrew it with the hammer? No - I’d pry it out or I’d whack it until it yielded. Removal of a screw is not a sufficient test to show that something is “screwdriver-like.”
A machine could take every message in every conversation ever had on Google Chat/Hangout, mine all of those conversations for similar messages, and use the subsequent messages as responses which are convincingly human (this is how most of the best “chatbots” currently work).
That requires very little “understanding” on the part of the machine, and the process is decisively non-human in nature. It’s difficult to make the case that it represents true machine “intelligence.”
But it might still pass the Turing test.
If the constraints on your machine become tighter (e.g. you’re limited in how much storage or processing power you have available), the algorithms and data structures must become more efficient and creative. Then you end up under the same conditions humans evolved from, and it’s much more likely a Turing-Test-passing machine will have many of the same characteristics as humans - which other answers have already mentioned.