One way to see if this can be physically possible is to think of deep blue. The first chess machine who beat Kasparov, the world's best player. Deep blue was able o beat the human thinking in a war game, however this machine was only programmed to think a pattern of algorithms to make the next move. In some sense deep blue would be a clear example of the beginning of a thinking machine, but be careful, not yet a rational thinking machine because deep blue is only programmed to recognize different sets of patterns and choose which move is the most probable move to win the entire game. For machines to feel is not entirely impossible if the robot has installed on their robotic arm a some sort of skin to sensor what it is around, and even better yet, this machine could be programmed to record all different senses with specific algorithms that have been programmed into a positronic brain. And again, is far beyond our time spam.
For machines to act like humans would be almost possible but not possible. the reason is because machines won't be able to understand what is love, or even understand the relationships of love and pain, or the relationship of dreams and life experiences. OUr brains are very similar to a computers pc, but the one from a pc is inorganic and not highly evolved, our brain can make calculations in 1 sec much more faster than what your regular pc would do. So for a machine to be like us would require for the machine to learn and be conscious, aware to be determined alive. The most popular believe between humans and terminator like machines is that many people think that machines don't have souls, that only humans have souls. However, incorrect.
The whole idea of thinking that humans have a soul is not scientifically proven, what we know as soul is our consciousness, so by the time a machine becomes self aware of its reality than we are screwed.
Machines and humans will be categorized differently, we would be monkeys and they would be cold calculating self aware machines who's superiority will exceed in many ways yet unknown. They could be highly intelligent, highly evolved, and just by thinking these things they might even reach a point were different patterns in their intelligence will make them act like we do, but never feel like we do because they are not humans. They are machines.
Today is not possible to make a robot like DATA because the technology is not there yet. But, one day we will have robots to do all basic things that men kind have done for years, like serving a drink for example replacing bar tenders, or buying and selling shares in the stock market, so a human can become rich.
I often think of FUTURAMA, where "Bender" a roboto from the future who likes to drink and slack off is almost the very same image of some human behaviors. A roboto with such qualities would replace dogs and cats as pets, this kind of robot is the friendly robot is the robot that everyone would like to have, but than we have SKYNET per example. A self aware highly intelligent machine who plans to kill the entire race. No one likes this robot because SKYNET would be the kind of robot that has a Napoleon complexity. So this AI still very robot, and has no idea of planetary diplomacy.
On the other hand we have ASIMO, the only human shape like robot who already does many things such as walk like humans on two legs, bring your food to your table, laugh at your jokes (even though they don't get it). So the thought that robots like humans are logically impossible is incorrect, ASIMO does many human like functions, because the robot was programmed to do these things, until a roboto doesn't become self aware of what is doing, we don't really know if the AI will have a human like personality, we don't even know if the robot will ever learn the set of human values and from there determine weather we should be killed or we should live. A robot is not highly intelligent from what is is today to do this kind of functions and reasoning.
No one likes the idea that some other intelligent "being" will one day walk with us. because we have been here for millions of years, but after all we are all basically made of the same ingredients, their made of metal and we are made of flesh, we all need atoms to be present in this world, they will age like us, they will rust and we will oxidize .
One great advantage is that they will never know the kinds of pleasure that we have unless these robot will be build to study the kind of pleasures that humans have, such as eating, sex, smelling, smiling, sarcastic jokes, affection. One example is that, if this kind of robot is the one from the movie Bicentennial Man, where a robot who functions as a servant seeking for human like experiences,than Andrew Martin, the name of the robot, could eventually be considered human after doing so many plastic surgeries to its body, with technology that might even help humans in the future to live longer. Although the committee who decided weather Andrew should be considered human or not had a rough time because Andrew still had a positronic brain rather than a normal organic brain, and "he had lived much more than any other human being". Ironic isn't it? at the end Andrew dies, his heart stops beating (yes he had a robotic heart) and his positronic brain stopped functioning (making it officially dead).
What ever the kind of robot we want to think of, if it is a killer robot, a robot with Napoleon Complexity, a robot with human like attitudes or a robot with an obsession with cosmetic surgery to become more human are still far from our time line, but this does not mean that they can't act like us or made the same mistakes that our race have done in the past.
