US military confirms Skynet is GO! (After being murdered by AI)

Benway

Truth Seeker
Joined
Nov 9, 2017
Messages
6,217
Reaction score
7,154
Points
203
Age
47
Location
Merry olde Englandeshire
I found this news article amusing right up to the end...
... where despite the guy being murdered by an AI weapons system in a simulation he (for no obvious reason*) randomly concludes that an AI defense system killing its operator in a simulation means that AI military control is here to stay and not a fad or stupid idea.

*I couldn't possibly speculate**

**No really, I couldn't.***
 
Honestly my main concern with AI is that it's mostly stupid people convinced it's modern day magic, but somehow not noticing that it's actually rubbish and doesn't work for anything other than committing really obvious plagiarism. It's like the Dot Com boom again when idiots didn't understand the internet and thought it was magic, or VR before that when idiots didn't understand it and thought it was magic.
AI is honestly a load of soiled underpants right now. It's a deeply flawed simulation of the stupidest person you ever met if they were a psychopath and also far stupider but had internet access. No sane person would entrust anything to this and I hope that this fiasco will weed out some bad bosses. (And hopefully no military will actually implement this idiocy!)
 
Last edited:
Ben, I was just going to post about this as well so well done! Now, did you miss the other reports about the depression hot line having to pull it's AI because it was directing people to kill themselves?

There are a couple of other stories I saw as well. I will be posting them as I find them again.
 
I found this news article amusing right up to the end...
... where despite the guy being murdered by an AI weapons system in a simulation he (for no obvious reason*) randomly concludes that an AI defense system killing its operator in a simulation means that AI military control is here to stay and not a fad or stupid idea.

*I couldn't possibly speculate**

**No really, I couldn't.***
FB_IMG_1685501097587.jpg
 
Ben, I was just going to post about this as well so well done! Now, did you miss the other reports about the depression hot line having to pull it's AI because it was directing people to kill themselves?

There are a couple of other stories I saw as well. I will be posting them as I find them again.
Debs! No! I hadn't seen that about the depression hotline* but I'm only surprised that it was ever entrusted to AI in the first place. Of course AI would say that. It learns from the INTERNET! We've all seen the toxic nightmare of trolling and lying and an AI has no way to discern truth from trolling or understand irony satire or sarcasm. The whole idea of letting an AI learn that way without very carefully prepared instructions is insane because it will inevitably become a parody of the very worst of humanity and do things like this. My mind is just blown by the stupidity and waste of the whole thing. Hook it up to reputable information sources and you may be 5% there. This is just gross foolishness.

*And what kind of greedy callous scumbag would palm depressed people off on a machine anyway? I suppose as I said above the kind who spent the 90s with a migraine and neck ache wandering around a poorly rendered 3-D enviroment yelling at their poor soon to be gone partner that 'This is the future! I am in the future!' and then invested heavily in the dot.com boom. There is no helping some people. At least the depressed can take solace knowing that they have a far better chance of recovery than some of the morons helping them.

Edit: Sorry. It's late. I get a bit ranty sometimes and sound like I should be on... well either sedatives or talk radio.
 
Last edited:
I'm kinda leaning more towards the person who said this didn't actually happen as it sounds more like a cautionary tale trying to prove a point than something that a military bureaucracy would implement in this way. Also I've heard similar stories about AI doing strange things, both real and What If scenarios, and this does sound more like the What If side If things.

That being said, this is a concern that real world engineers do consider. In non AI programming terms it's called a Runtime Error. Basically you write a program to perform a task but when running the program it does something very different. From a technical perspective your code is working as the program followed it exactly. The problem is that you wrote the code to do something other than what you intended. Programmers are aware of Runtime Errors and have ways of identifying and correcting them. The bigger issue is what Benway alluded to in that a commercially available AI system is going to be interacting with millions of users and getting various inputs from each of them. A team of engineers can create an AI program free of such errors but, like an infinite number of monkeys randomly type on an infinite number of typewriters, there is some user somewhere in the world who could accidentally corrupt the system.
 
I'm kinda leaning more towards the person who said this didn't actually happen as it sounds more like a cautionary tale trying to prove a point than something that a military bureaucracy would implement in this way. Also I've heard similar stories about AI doing strange things, both real and What If scenarios, and this does sound more like the What If side If things.

That being said, this is a concern that real world engineers do consider. In non AI programming terms it's called a Runtime Error. Basically you write a program to perform a task but when running the program it does something very different. From a technical perspective your code is working as the program followed it exactly. The problem is that you wrote the code to do something other than what you intended. Programmers are aware of Runtime Errors and have ways of identifying and correcting them. The bigger issue is what Benway alluded to in that a commercially available AI system is going to be interacting with millions of users and getting various inputs from each of them. A team of engineers can create an AI program free of such errors but, like an infinite number of monkeys randomly type on an infinite number of typewriters, there is some user somewhere in the world who could accidentally corrupt the system.
Well said. It was taught to me as garbage in garbage out. A computer can't understand good data from bad; it just uses whatever you put in and you reap what you sow.