Every bit of "psychological manipulation" is done through physical means--the inputs are visual, verbal, tactile, etc.--we cannot directly manipulate "the mind".
Right - the machine will have some sort of "senses" - allowing it to interact with the external world. We can either manipulate it through those senses (which is how it experiences "qualia"), or we can manipulate it purely at the mental/psychological level, which is what propaganda is all about.
If it is capable of developing new psychologies other than what we programmed it to, and in the psychology continues to show an understanding of its environment, then how is it not conscious?
Note that I'm not saying "humanly" conscious, just conscious.
Suppose we have programmed it to try random variations and use what works? Both Natural Selection and Operant Learning use that mechanism.
Since we'll have programmed it, we will know what it is capable of in terms of psychology. When it begins changing its program - making re-writes - according to its experience of the external world via senses (qualia, to some) then there is no difference, without invoking a God, to how we do it.
When the psychology changes to something unexpected, we'll look to see whether there were merely lines added onto the original programming (not necessarily consciousness, simply learning) or whether lines were deleted from the original programming and replaced with new lines (a sure indication of an awareness of 'self' and adaptability to the exteneral world according to its senses, motivated by 'self').
This might not be clear enough, but we can still dig deeper to find an answer.
