Sherry Turkle: Alone Together MARCH/APRIL 2011 95 systems, the more their users might assume intelligence beyond the sys- tem’s capacity. If systems have built- in learning mechanisms that enhance the impression of intelligence, humans could imbue their robotic helpers with even more capability. I’m reminded of my own frailty: most big mistakes I make start with an assumption that was wrong, leading to a cascade of consequences. I ’m not sure where I’m going with this, except to say that I’ve come away from reading Turkle’s book with a profound respect for the hazards of creating simulations of viviform creatures. In her stories, we learn that adults and children felt rejected when the simulacra weren’t responsive, or imbued them with properties they couldn’t possi- bly have. Even the most primitive of simulations such as the famous Eliza program developed in the 1960s by Joseph Weizenbaum led users to seek privacy for intimate “discussions” with this text-based robot (http:// en.wikipedia.org/wiki/ELIZA). Turkle also recounts stories of apparent beneft in which the robots provide companionship and a sense of worth to lonely, aging users. The conundrum of beneft through empty fakery left me unsure how to think about our present-day capac- ity to create such artifcial creatures. In any event, I feel far more sensi- tized to the potential side effects of attempting to create convincing simulations of familiar life forms. If there are lessons here, at least one must be the need to consider the consequences on the human side of the equation of emulating natural interactions with programmable, but essentially very limited, robots. Vinton G. Cerf is vice president and chief Internet evangelist at Google. Contact him at vint@google.com. cont. from p. 96