3

After watching the latest iteration of the Terminator series I am struck with two observations: offsite backup is important, and couldn't they have avoided the whole rogue AI with proper testing?

As a software developer I know that testing an application is part of the software development process.

Could the nuclear war created by Skynet have been averted with robust testing of the AI?

closed as off-topic by user14111, Ward, phantom42, Often Right, Joe L. Jul 16 '15 at 11:16

This question appears to be off-topic. The users who voted to close gave this specific reason:

  • "Questions seeking scientific solutions or explanations are off-topic unless they relate directly to a cited work of fiction. There are a number of other Stack Exchange sites dedicated to answering questions on non-fictional sciences." – user14111, Ward, Often Right
If this question can be reworded to fit the rules in the help center, please edit the question.

  • 3
    I remember in T3 the Arnienator said "Judgement day is inevitable". So, no. Remember also that in the same movie, John Connor realises that "Skynet is the virus", so essentially it had infiltrated the entire Internet before even becoming "Self-Aware". See How did Skynet screw the world even before its activation? – Möoz Jul 16 '15 at 2:52
  • Also, as you might remember famously that "the loveliest trick of the Devil is to persuade you that he does not exist!"*-Charles Boudier* – Möoz Jul 16 '15 at 2:57
  • 1
    How do you test for possible future development of sentience and malice? – user16696 Jul 16 '15 at 5:44
  • Good news, boss! Skynet works great. It fulfills all the project requirements and it's totally reliable. I just want to run one final test: Give it simulated feedback for a while and don't actually put it in control of anything, in case it becomes sentient and tries to destroy humanity. Another $4 billion and 10 years ought to do it. You're cool with that, right boss? – Royal Canadian Bandit Jul 16 '15 at 9:21
5

As a software developer, you should also know that:

  • Nobody can ever write ALL the tests possible.

  • You typically tests for things you know/predict will go wrong, largely based on past experience. Skynet turning into homicidal warmanger isn't a likely scenario to have crossed an average programmer's mind. Every geek saw "WarGames" and thinks that all AIs are benevolent and don't wanna wage evil war.

  • It's really hard to test complex interactions of complicated software.

    Such as having your tested AI being infected by a new, unknown, unkillable distributed virus, even if you did somehow manage to predict you need that test case because you just don't identify with Matthew Broderick.

  • 1
    I'm also going to assume that Dennis Nedry was Cyberdyne's QA guy. – Liesmith Jul 16 '15 at 8:16
  • @Liesmith - Nedry's software ran pretty well, till he intentionally sabotaged it. As usual, the best method of hacking is socially-engineered slipping a proper dude a proper-sized stack of $100 bills at a proper time – DVK-on-Ahch-To Jul 16 '15 at 15:09
  • There are other methods. – kylie.a Aug 3 '15 at 12:17

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