What AI is not

We shouldn’t confuse the images we see in science fiction with the reality of AI.

Killer robots are not coming to get us

Popular culture makes it very hard for us to think about AI in a clear-sighted way. For decades it has filled our imaginations with all manner of killer robots, devious androids, and out-of-control software programs coldly plotting to enslave or exterminate their human makers. These images have consequences that reach beyond pop culture. They fuel debates in university philosophy departments about how to prevent hypothetical superintelligent AIs from one day taking over the world.

We should rid ourselves once and for all of these fantasy ideas about AI. They are nothing like the reality. Today engineers are working on applying AI’s pattern recognition and prediction skills to solve problems that, although limited, are of great practical value to us humans.

The super-intelligent AI of science fiction that can think and understand the world as well as we do (or even better), that can harbor intentions that are “good” or “evil,” that can perhaps take control of the world for our own good or even plot to overthrow us, that AI simply does not exist. Maybe it will exist one day in a distant future that no one can foresee today… or maybe not. But what is certain is that this AI of science fiction is not even on the drawing boards of today’s AI engineers.


What AI is not:

  • A super robot that thinks like a human

  • Software that has a deep understanding of the world based on years of lived experience

  • Something that understands good and evil (beyond what we tell it)