
Started By
Message
The future of artificial intelligence...
Posted on 3/25/16 at 9:40 am
Posted on 3/25/16 at 9:40 am
Some bright bulb at Microsoft Research had the clever idea of turning a machine learning program loose on Twitter yesterday to learn how humans interact with each other. Humans, predictably, interacted terribly.
Within 24 hours, the Twitter bot @TayandYou had turned spectacularly ugly in the way only the internet can turn things ugly, spewing racism and hate at Twitter uses in a series of horrifying tweets. Most of them have already been deleted, perhaps in a bid to ensure that any future artificial intelligence (AI) won't have evidence of what horrible bastards humans were back before AI took over the world. Or maybe Microsoft was simply trying to avoid bad PR. Regardless, there are screenshots.
LINK
Within 24 hours, the Twitter bot @TayandYou had turned spectacularly ugly in the way only the internet can turn things ugly, spewing racism and hate at Twitter uses in a series of horrifying tweets. Most of them have already been deleted, perhaps in a bid to ensure that any future artificial intelligence (AI) won't have evidence of what horrible bastards humans were back before AI took over the world. Or maybe Microsoft was simply trying to avoid bad PR. Regardless, there are screenshots.
LINK
Posted on 3/25/16 at 9:57 am to Rebel Land Shark
This. If this is not a sign that AI will get out of hand than nothing is. Though it was intended for good, some will find a way to use it for bad
Posted on 3/25/16 at 10:08 am to Alahunter
I thought I've read all the leaning it does is just recycled human tweets.
Not this time Skynet
Not this time Skynet
Posted on 3/25/16 at 10:31 am to Alahunter
More 4chan shenanigans. Theyre always good for a chuckle.
Posted on 3/25/16 at 10:34 am to Alahunter
This was either on the OT or Poli Board yesterday; I compared it to the Ask Jameis thing that Fl St did a few yrs ago as in 'what could possibly go wrong'? 
Posted on 3/25/16 at 10:35 am to Alahunter
quote:
it isn't all that far off from the way very young human or animal babies learn by watching and emulating adult behavior.
AI will at first be a mirror of humanity. We shouldn't be alarmed by what this AI did. Rather, we should be upset by what it saw and mirrored back to us.
quote:
First off, it's likely that the internet is already changing human patterns of thinking and behavior, drawing people away from deep thinking. According to a paper published last year in Neuroscientist, the internet is training people to "gravitate toward 'shallow' information processing behaviors characterized by rapid attention shifting and reduced deliberations." The internet does not catch people at their best.
How many times have you seen "TL;DR" as a retort to someone's attempt to share information in more than a few sentences? I expect a troll(s) to respond accordingly to this post.
A lot of forums devolve into quips and insults.
quote:
But the bigger (and related) problem, already well known to anyone who has spent more than 40 seconds online, is trolls. The internet turns out to be a wonderland for people with a cluster of really negative personality traits called the "Dark Tetrad": narcissism, psychopathy, sadism, and ruthless self-interest (Machiavellianism).
tRant is infested with trolls. There are enough good posters on the SEC Rant to stay, however.
quote:
while the Microsoft developers intended for Tay to learn about communication from humans, it ended up learning about being an a-hole from a pack of sadistic psychopaths.
Psychological experiments have consistently shown that humans will inflict pain on other humans at a startlingly high percentage, even if there is no immediate reward for doing it. In real life this is borne out by examples such as Nazi Germany and even Americans' historical resistance to expand civil rights to minorities that they consider to be inferior.
AI will be fine when it grows enough to become logical. Humans tend to be illogical.
This post was edited on 3/25/16 at 10:37 am

Popular
Back to top
5









