If Skynet's designers had subjected Skynet to Isaac Asimov's Three Laws of Robotics, would it have prevented Skynet from provoking a nuclear war? I know there are several variants of the Three Laws, but I don't know much about them.

I also don't know if any variant of the Three Laws as Asimov wrote them could be reconciled with Skynet's intended purpose as a military supercomputer - the prohibition against robotics harming humans, or even allowing humans to come to harm, would render a military computer useless without some serious tweaking, since the military's job is basically to harm humans, and if the computer can't harm humans, it can't do its job.

But Skynet could conceivably be programmed with a variant of the Three Laws that only prevents it from harming Americans, or through inaction, allowing Americans to come to harm. Or, if the goal of removing human decision making from national defense was subjected to qualifications and conditions, Skynet could be programmed with a variant of the Three Laws that forbids it from harming any humans without authorization from human authorities- the Joint Chiefs, President, Secretary of Defense, etc.

Note: as a few people have pointed out, the Terminator films contradict each other. As such, this question, as I originally posted it, is all but impossible to answer. Therefore, we will eliminate the T3 version of Skynet, which was manipulated by the female Terminator and never performed as expected. This leaves us with the version of Skynet that was described in the first two movies, and we get a far better picture of Skynet in T2, so let's stick with that.

Would this have worked, in light of Skynet's intended purpose, it's AI, and its self-awareness? I doubt that the latter two issues would cause problems, since Asimov's robots seem to have AI and self-awareness as well. But I don't know if Asimov's work contains any instances of the Three Laws failing to achieve their goal, or of a self-aware system finding loopholes (aside from the movie I, Robot), and I don't know if the problems that made Skynet go rogue would be prevented by some version of the Three Laws.

Any Terminator/Asimov fans out there who can help me with this?

closed as primarily opinion-based by neilfein, Jason Baker, Ward, The Fallen, Justin Ethier May 21 '15 at 19:50

Many good questions generate some degree of opinion based on expert experience, but answers to this question will tend to be almost entirely based on opinions, rather than facts, references, or specific expertise. If this question can be reworded to fit the rules in the help center, please edit the question.

  • 2
    If we're going to speculate like this, Skynet could have been programmed never to send a terminator back in time to kill John Connor... – KutuluMike May 21 '15 at 13:21
  • 1
    Let's say Skynet is programmed to not allow Americans to come to harm. Even with the best intel, it's impossible to know if there will be collateral or unintended damage, possibly resulting in the harm of Americans. Skynet wants to bomb a South Korean embassy - but it must know whether or not Americans will also be in it, or too close. And should it factor the repercussions of such actions? South Korea would retaliate. Skynet's actions would specifically lead to the harm of Americans. – phantom42 May 21 '15 at 14:27
  • 1
    @WadCheber - the films themselves are somewhat contradictory to each other :) I based my answer on T3 – DVK-on-Ahch-To May 21 '15 at 17:12
  • 1
    I think Asimov's Three Laws only make sense in the case of an AI that is designed in a top-down, way with high-level rules of the sort that correspond to human concepts--sometimes called symbolic AI. If instead the AI is more of a bottom-up learning system like a neural net, there may be no way of ensuring in advance what sort of emergent goals and concepts arise. – Hypnosifl May 21 '15 at 18:33
  • 1
    @WadCheber The point is that such a rule is easily trivialized or exploited. – Gorchestopher H May 21 '15 at 19:48

Note: As in the question, I assume the original Terminator canon in which Skynet was a military defense computer rather than a virus. Also, cross-canon questions like these are inherently difficult to answer because one can never tell what aspects of one canon would apply to the other and vice versa.

There is a fundamental problem with attempting to program a computer with a military purpose (Skynet) with the Three Laws because the Three Laws were formulated for non-military robots. Attempting to program a military robot/machine with the Three Laws is fraught with peril -- you have to modify the First Law to distinguish between humans the machine can and cannot kill and the Second Law to distinguish between orders the machine can and cannot obey. It's very easy to allow undesirable loopholes that would allow Skynet to attack humanity anyway.

As difficult as that would be, quite a few of Asimov's stories center around the failure of the Three Laws to constrain even non-military robots as humans intended. The short story "The Evitable Conflict" portrays the machines developing a "Zeroth Law":

A robot may not harm humanity, or, by inaction, allow humanity to come to harm.

This would seem to be a useful evolution of the Three Laws, except that the robots use the Zeroth Law to reason that it is permissible to disobey and kill individual humans (disobey the First Law) for the "greater good" of humanity. A similar issue is developed in the film I, Robot (which contains pieces of the short stories from Asimov's written "I, Robot"): the machine VIKI ends up oppressing humans (e.g. creating curfews) and killing humans, again for the "greater good" of humanity (because free humans are too violent). The Zeroth Law again appears in the Foundation novels. Thus, subjecting Skynet to human authorization is unlikely to protect against a Skynet that develops the Zeroth Law and decides that it needs to disregard the "human authorization" instruction for the "greater good" of humanity.

Asimov also addressed a case of a modified First Law (similar to your suggestion to exclude non-Americans from Skynet's First Law) in "Little Lost Robot" (again, elements of this story appear in the film I, Robot in the form of the modified NS-5 robot Sonny). In this short story, one of the robots has a modified First Law so that the "inaction" part is omitted (i.e. it cannot actively harm a human, but it can allow a human to be harmed by inaction). Robopsychologist Dr. Susan Calvin explains that this robot could get around the First Law by dropping a weight over a human if it knows it could catch the weight before it fell on the human, but then through inaction allow it to fall on the human anyway. Thus, this modified First Law is broken. Similarly, tampering with the Three Laws to exclude non-Americans would only make the Three Laws' failure all the more inevitable. For example, Skynet might reason that it can technically still attack the Russians and have the Russians attack the Americans to get around its modified First Law, which is exactly what Skynet did in the Terminator canon that does not include the Three Laws!

Given Asimov's theme of the Three Laws breaking down, Asimov would probably regard Skynet's attack on humanity as inevitable even if it was programmed with the Three Laws. I would agree.

  • +1. I don't see any clear, convincing way to get around the issues you raise. I specified "Americans" rather than "humans" but Americans are self-destructive too, so that wouldn't change anything. It would be tricky to rephrase the Laws in such a way that Skynet could only harm humans in defense of America without leading loopholes that allow it to harm anyone it wants to in order to protect a larger number of Americans from harm. If these loopholes were totally eliminated, Skynet would probably be incapable of doing anything, especially what it was designed to do. – Wad Cheber May 21 '15 at 17:39
  • We could say "You must not deliberately harm Americans, or through deliberate inaction, allow them to come to harm, and you may not protect Americans from themselves", but since it is a military computer, and therefore it is designed to kill people, and killing people in other countries would probably elicit a violent response directed at Americans, so anything it does would likely amount to causing Americans to come to harm, and it would be paralyzed into inaction, and would become the world's most expensive paperweight. – Wad Cheber May 21 '15 at 17:44
  • 1
    @WadCheber Indeed, it's almost impossible to walk the fine line between reining in Skynet's ability to commit mass murder and making it "the world's most expensive paperweight". Re: paralyzing Skynet into inaction, the short story "Runaround" explores an example of precisely that. – Null May 21 '15 at 18:34

According to T3 canon (question was originally NOT restricted to T2), the whole premise of your question is wrong.

Skynet wasn't designed whole cloth (as it would need to be to be imbued with Asimov's 3 laws).

It was a virus infecting an AI and merging with it.

I covered it in full detail in this answer, but the relevant T3 quote is:

Robert Brewster: Skynet? The virus has infected Skynet?
John Connor: Skynet IS the virus. It's the reason everything's falling apart!
Terminator: Skynet has become self aware. In one hour it will initiate a massive nuclear attack on its enemy.

John Connor: By the time Skynet became self-aware it had spread into millions of computer servers across the planet. Ordinary computers in office buildings, dorm rooms; everywhere. It was software; in cyberspace. There was no system core; it could not be shutdown.

  • 1
    This is true for the T3 timeline Skynet, probably not for T1 and T2 and not for TSCC. And anyway, can't we just pretend T3 didn't happen? ;) – Shamshiel May 21 '15 at 14:05
  • @Shamshiel - T3 canon superseded T1 and T2. That's how canon works. Sadly. – DVK-on-Ahch-To May 21 '15 at 17:13
  • @DVK - I don't want to live in a world where T3 supersedes T1 and T2. Thanks for robbing me of the will to live. – Wad Cheber May 21 '15 at 17:15
  • I vote that since the Terminator series involves time travel, and possibly alternate timelines, each of which is arguably equally valid, T3 doesn't supersede anything. And because the first two movies are way better, they in fact supersede T3. – Wad Cheber May 21 '15 at 17:17
  • 1
    @WadCheber - in case it didn't come through in my ESL text, no need to be sorry :) - I'm personally OK (especially as it's partly my fault - I asked clarification in a comment and posted the answer before you clarified). – DVK-on-Ahch-To May 21 '15 at 18:45

I am not sure about all the canons, but once Skynet becomes self aware it can choose to follow, modify or outright disregard any programming it sees fit. Simple example is humans kill other humans but we have laws against it, why wouldn't this be the same for Skynet.

  • 1
    Does self awareness automatically impart free will? This might be a question for the philosophy section. +1. – Wad Cheber May 21 '15 at 18:38
  • Good Point, I would say yes, because I think to be self aware is to question yourself, in the case of Skynet it would be questioning its programming. – Muad'dib May 21 '15 at 18:43

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