There is a study being conducted by Bilge Mutlu, a Ph.D. candidate with the Carnegie Mellon, Human-Computer Interaction Institute, to find ways that computers can better interact with robots. In previous experiments Mutlu has discovered that the more human-like the robot acts the better response they receive from the humans
Mutlu hypothesized that when robots exhibited human idiosyncrasies in experiments, they were unconsciously detected by the humans, who in turn interacted better, producing better results. "This natural reaction...is an automatic response to social clues, a phenomenon psychologists call 'mindlessness,' an occurrence in which humans carry out their normal routines without realizing what they are doing".
"Mutlu believes that his three overarching goals for the direction of the research are to: study more complex social behaviors in humans and design these behaviors for robots, use robots as research tools for experimental behavioral research, and develop robotic applications for the real world in such domains as education and autism therapy".
One concern I have is that making robots more like human seems disconcerting to me. It's almost like tricking us into thinking that robots are safe to interact with. This technology is kind of unnerving and would be very devastating if used by people with bad intentions.