One step closer to Skynet?

[Daird]

http://gizmodo.com/this-is-the-first-computer-in-history-to-have-passed-th-1587780232/all

Computer scientists in Russia have apparently designed a program that can mimic a thirteen-year-old in a chat room. 33% of the study participants were duped into thinking the program was human, edging out the standard of 30% used for the Turing Test. Many computer experts thought this was possible, but it doesn't seem like any of them saw it coming anywhere close to now. Does anyone here think we might have underestimated our own technological progress all along?

 

[dutchfire]

Moderator Action: You've given a Red Diamond designation to this thread. This means that discussion in this thread will be held to a higher standard than usual, and that includes the Opening Post. Could you please expand your opening post to explain what exactly is in the link and what you want to discuss about it?

 

[Kozmos]

Nope.

"Eugene" was created by a team based in Russia, and passed test the organized by the University of Reading just barely, by duping 33 percent of the judges. It should also be noted that successfully pretending to be a 13-year-old boy for whom English is a second language ain't exactly Hal 9000.

It means basically it's like all the other chatting software that spouts nonsensical gibberish at times.

 

[Kyriakos]

The turing test (and similar) is not something which determines the existence of actual AI anyway. It merely determines if people can be fooled into thinking there is a human on the other side. That doesn't mean much; in some strategy games, with little experience, you might think an online 'player' was not a fill-in AI either.

More importantly: this makes the computer no closer to having an actual intelligence than a dog painted decently and permanently blue is to an actually naturally occuring blue dog.

 

[Bootstoots]

Sounds like the computer managed to convince 33% of a panel of judges that it's a 13-year-old non-native English speaker. Maybe that counts as a technical pass, but I don't think it's as high of a standard as some might make it out to be.

I wouldn't be surprised if Big Data applied to language results in computers that pass more demanding versions of the Turing test by analyzing billions of real-life human chats and coming up with a machine that gives a believable response to most inputs (and a bizarre one no more often than average people do). This wouldn't mean that AI will take over the world, but it could mean that advertising bots eventually become hard to distinguish from real people all over the Internet. In other words, more work for moderators.

 

[Terxpahseyton]

Bootstoots seems to sum it all up.

Perhaps in the far future, bots can socialize for us on whatsapp or facebook if we don't feel like it but don't want to be rude (so we will secretly be even ruder)

 

[Kyriakos]

A working clone would do well as an alibi

 

[Kozmos]

Bootstoots seems to sum it all up.

Perhaps in the far future, bots can socialize for us on whatsapp or facebook if we don't feel like it but don't want to be rude (so we will secretly be even ruder)

Which is not that bad of an idea, scan yourself and have your own digital version serve as your assistant for all electronic matters. We could even have digitized tombstones so people can talk to you after your fleshy bits are dead. Now you never have to let go!

 

[CavLancer]

It seems to me that most 13 year old boys are more than 33% gibberish so I'm not sure how great an accomplishment this is.

 

[El_Machinae]

Perhaps in the far future, bots can socialize for us on whatsapp or facebook if we don't feel like it but don't want to be rude (so we will secretly be even ruder)

I have a more immediate need than this. My sister is severely mentally handicapped. Loneliness and boredom are her biggest issues. Having something to interact with would help her so, so much.