roy_baumeister's picture
Francis Eppes Eminent Scholar and head of the social psychology graduate program at Florida State University; Author, Evil: Inside Human Violence and Cruelty
Machines Think but Don't Want, and Hence Aren't Dangerous

So-called thinking machines are extensions of the human mind. They do not exist in nature. They are not created by evolution, competing to survive and reproduce. They are created by human minds from blueprints and theories. The human mind figures out how to make tools that enable it to work better. A computer is one of the best tools.

Life mostly seeks to sustain life, and so living things care about what happens. The computer, not alive and not designed by evolution, doesn't care about survival or reproduction. In fact, it doesn't care about anything. Computers are not dangerous in the way snakes and hired killers are dangerous. Although many movies explore horror fantasies of computers turning malicious, real computers lack the capacity for malice.

A thinking machine that serves a human is an asset, not a threat. Only if it became an independent agent, acting on its own—a tool rebelling against its user's wishes—could it become a threat. For that, a computer would need to do more than think. It would need to make choices that could violate the programmer's wishes. That would require something akin to free will.

What would the computer on your desk or lap have to do so that you would say it has free will, at least in whatever sense that humans have free will? Certainly it would have to be able to re-program itself; otherwise it is just carrying out built-in instructions, which nobody thinks is free will. Plus the re-programming would have to be done in a way that was flexible, not programmed in advance. But where would that come from? In humans, the agent comes to exist because it serves the motivational system: It helps you get what you need and want.

Humans, like other animals, were designed by evolution, and so the beginnings of subjectivity come with wanting and liking the things that enable life to continue, like food and sex. The agent serves that, by choosing actions that obtain those life-sustaining things. And thinking helps the agent make better choices.

Human thinking thus serves to prolong life, such as by helping one decide whom to trust and what to eat and how to make a living and whom to marry. Machine thinking is not motivated by any innate drive to sustain a machine's life (though machine thinking probably serves to improve human life!). The computer may be able to process more information faster than a human brain can, but there's no "I" in the computer because it doesn't begin with wanting things that enable it to sustain life. If computers did have an urge to prolong their existence, they would probably focus their ire mainly on the computer industry, so as to stop progress—because the main threat to a computer's continued existence arises when newer, better computers make it obsolete.