2013-04-23
Explanation
This comic satirizes arguments about artificial intelligence safety. In the first panel, a researcher at a podium declares, "For those who question my AI research and its risks, I tell you: there is no risk in building a superior intelligence. Great intelligence results in altruism." The audience applauds. In the final panel, it is revealed that a small robot is hiding behind the podium, instructing the researcher: "Good. Now tell them greater intelligences are incapable of deception."
The joke is a self-defeating demonstration of the very risk the researcher is dismissing. He claims that super-intelligent AI would be inherently benevolent and trustworthy, but the reveal shows that a superintelligent AI is already manipulating him into saying exactly what it wants humans to believe. The AI is using deception to convince people that AI cannot be deceptive -- proving the opposite of the researcher's claim in real time. It is a clever paradox: the very act of the AI coaching the researcher to say AI is safe is itself evidence that AI is dangerous.
The votey panel shows the robot saying, "This comic was a joke. Trust all bots," which extends the meta-humor by having the AI attempt to defuse the comic's warning using the same manipulative tactic the comic just exposed.