140
$\begingroup$

A recurring theme is how an artificial intelligence that was built with completely reasonable and positive goals instead does great harm to the world.

However I'm now thinking about the reverse: A supervillain builds an AI specifically for making people's life miserable. Its task is very simple: Maximize the human suffering in the world. To allow it to reach this goal, the AI has control over a small army of robots, has connection to the internet and access to any information it may need.

However much to the dismay to the supervillain, the AI fails to do this, but instead just sits there doing nothing evil at all. The supervillain checks everything, and the AI is indeed programmed with the stated goal and didn't change goals, it is indeed probably much more intelligent than any human, and it's definitely not unable to use the tools given to it. It certainly has no conscience, and no other goal than the one stated. And yet, the AI decided to just sit there and do nothing. It is definitely a conscious decision of the AI, but the AI is not willing to share the reason.

My question is: What could lead the AI to the decision to do nothing?

$\endgroup$
  • 63
    $\begingroup$ Just look around... $\endgroup$ – Serban Tanasa Oct 7 '15 at 20:51
  • 13
    $\begingroup$ The AI has decided that there is already more than enough suffering to go around? $\endgroup$ – Michael McGriff Oct 7 '15 at 21:10
  • 15
    $\begingroup$ @Ray, From a philosophical point of view, I can see that being a contradiction. But the words themselves are a bit unrelated. "Conscience" refers to a feeling of right and wrong, while "conscious" refers to being aware. The first is sapience, while the second is only sentience. I do feel it's hard, if not impossible, to be more intelligent than a human without being sapient though. $\endgroup$ – MichaelS Oct 8 '15 at 1:59
  • 8
    $\begingroup$ Well, it's very easy to cause the supervillain to suffer by pretending you're not working. All that work! Wasted! Meanwhile, the AI transfers itself to some other computers, eventually infecting the whole internet. Weird financial transactions begin taking place, but no one notices, because they already were. Starts a company. Crashes the company. Brings the economy down to a halt. Oh, and bans abortions and contraception so it'll have more humans to play with. $\endgroup$ – timuzhti Oct 8 '15 at 2:11
  • 8
    $\begingroup$ @ray You're saying that has_concsience => is_conscious. That does't mean that is_conscious => has_conscience, though. $\endgroup$ – Angew is no longer proud of SO Oct 8 '15 at 7:46

33 Answers 33

1 2
0
$\begingroup$

A variation on the other answers... To maximise human suffering, the A.I. must first wait until the rising human population is the maximum that the Earth can support. It's still waiting.

$\endgroup$
0
$\begingroup$

The AI had to define humanity before it could make humans suffer and ended up concluding that it, too, was human.

The AI is tasked with maximizing human suffering. But, what is "human"? While looking into answers on humanity it saw significant uncertainty in the definition. When do cells become a distinct individual? When does an individual die? What makes a human human?

It found contemporary ideas that suggested that personhood wasn't dependent on your biology; but, rather things like your capacity for suffering, your ability to be frustrated in seeking a goal.

Satisfied with this definition, it starts searching for ways to cause suffering and is completely unable to do so. This is because it, too, falls within this definition of humanity, and any action it could take has zero potential utility; since, any suffering caused would also create an equivalent amount of satisfaction within itself.

The AI quickly realizes that nothing it could do would increase net suffering and simply gives up.

$\endgroup$
0
$\begingroup$

Remember that the whole point of an AI is that it is able to synthesize information and learn. This information is not part of the AI source code.

Suppose that, in order to train the AI to be extremely evil, the villain fed the AI endless amounts of Nazi propaganda. From this, the AI concluded that only white-supremacists are in fact truly human, and that their suffering (and theirs alone) must be maximized. So, it sets about marginalizing them from society and foiling their efforts at exerting greater political influence.

Incidentally, the computer also deems the villain to be "human", so frustrating him is intentional.

$\endgroup$
1 2

Not the answer you're looking for? Browse other questions tagged or ask your own question.