If solipsism is to be believed, then the machine can think only what you, yourself thinks. There is an irony here, because throughout history we have programmed the computer and therefore decide what it does and doesn't "think" so the idea of computers being able to think has been solipsistic from the beggining. However, the question begins to hinge as computers progress on whether you believe in solipsism, because soon the computers will be programming themselves, and at that point it is no longer so clear. It is sad that the idea of whether or not computers can become sentient may be based on something as weak foundationalism.

References: solipsism, epistemology, foundationalism