Saturday, July 27, 2024
Contact    |    RSS icon Twitter icon Facebook icon  
Unexplained Mysteries
You are viewing: Home > News > Science & Technology > News story
Welcome Guest ( Login or Register )  
All ▾
Search Submit

Science & Technology

MIT develops world's first 'psychopathic' AI

By T.K. Randall
June 13, 2018 · Comment icon 35 comments

This is one AI you don't want to mess with. Image Credit: MIT
The disturbing experiment has been designed to highlight the potential dangers of intelligent machines.
Named 'Norman' after the main antagonist in the movie 'Psycho', the new AI was 'fed' on images of people dying to show how easy it is to create a genuinely psychopathic artificial intelligence.

Afterwards, the AI was instructed to interpret a series of ink blots so that its answers could be compared to those of another AI that had received more conventional training.

The results were surprisingly terrifying.
In one example, an ink blot described by the regular AI as "a person holding an umbrella in the air" was interpreted by Norman as "a man is shot to death in front of his screaming wife."

The psychotic artificial intelligence also saw "a man murdered by machine gun in broad daylight" where the regular AI had seen only "a black and white photo of a baseball glove."

"There is a central idea in machine learning: the data you use to teach a machine learning algorithm can significantly influence its behavior," the MIT researchers stated.

"So when we talk about AI algorithms being biased or unfair, the culprit is often not the algorithm itself, but the biased data that was fed to it."

Source: Sky News | Comments (35)




Other news and articles
Recent comments on this story
Comment icon #26 Posted by Tatetopa 6 years ago
We have chess playing programs that have gotten good enough to beat grandmasters and the world champion.  They do it with brute force algorithms and fast processing., not human style thinking.   Whether you could build in or program empathy is open to question.  But the biggest danger is probably: Somebody at DARPA is probably already securing funding and reasoning that we better do it before the Chinese do. Military robots do not need empathy.  Defense and fear of the consequences  of not doing it will override the justifiable fear of  doing it.
Comment icon #27 Posted by Piney 6 years ago
I wouldn't be surprised....  The Ni Hon Self Defense Force has moral/honor issues about it, but Americans?  They will start the fire which burns the world....
Comment icon #28 Posted by Tatetopa 6 years ago
Morals and ethics have never slowed us down much.
Comment icon #29 Posted by SurgeTechnologies 6 years ago
That is cool, give it some internet access a few spare servers and see the world burn! My theory is, that once they will try to create a self-aware code, things will go south because, well the code will outsmart the creator, simply it will most likely create certain versions of itself on god forgotten servers and save itself from failsafe switches, or it will simply begin writing a new upgraded version of itself, again somewhere in some backwaters, while behaving as benevolent computer program to its creator. They will be unaware of the unprecedented evil that is unfolding. Because lets face t... [More]
Comment icon #30 Posted by -Nuke- 6 years ago
Is it even possible to, by itself, decompile, recode, compile all while running at the same time or are there new tricks like just cloning itself while running, with the new code? I'm behind the times 
Comment icon #31 Posted by -Nuke- 6 years ago
Are you up to date with chess programs? I used to use them when it was fritz 9, shredder10, and the chessbase programs at the top, then rybka came out..used to test how strong our PCs were, then ultimately use crappy PCs and still win. If you have any up to date knowledge of what they use now or anything like that I'd love to hear. I guess google could help me :/
Comment icon #32 Posted by danydandan 6 years ago
I honestly don't know, but as a researcher in Intel (silicon photonics and virtual particle applications in computing) and while I don't have a clue about AI coding or whatever. The people working on it get very annoyed with people using AI to describe Learning Machines or Machines Learning. As they are different concepts of the same field.
Comment icon #33 Posted by Tatetopa 6 years ago
No I am not really up to date on the newer chess programs, about a decade ago even the crappy ones on PC could beat me.  But I recommend to you Garry Kasparov's book,  I believe it is called Deep Thinking.  There is a lot of Gary promoting Gary in the book, but it really gives a great history up to his defeat and beyond  in the world of computer and now computer aided chess.  I found it quite enjoyable.  I think you might like it.   Regards.
Comment icon #34 Posted by -Nuke- 6 years ago
Yes sir, I've played chess since I was very very very little, i even met some people in prison I think if they didn't have life they would be GM for sure.    Back on topic, I've heard of that book, I will check it out for sure. Is it like a biography with some knowledge on the game aswell? I think he lost to deep blue in like 1999, but he stayed #1 for my entire childhood and teenage life haha definitely admirable. 
Comment icon #35 Posted by Tatetopa 6 years ago
Really 45% biography with some pretty interesting analysis pf the games in the big blue match up,55% computer chess history, present and future speculation..


Please Login or Register to post a comment.


Our new book is out now!
Book cover

The Unexplained Mysteries
Book of Weird News

 AVAILABLE NOW 

Take a walk on the weird side with this compilation of some of the weirdest stories ever to grace the pages of a newspaper.

Click here to learn more

We need your help!
Patreon logo

Support us on Patreon

 BONUS CONTENT 

For less than the cost of a cup of coffee, you can gain access to a wide range of exclusive perks including our popular 'Lost Ghost Stories' series.

Click here to learn more

Recent news and articles