Pages:
Author

Topic: Turing Test Pass Claimed at Royal Society Event (Read 1088 times)

sr. member
Activity: 770
Merit: 250
Well, that bot was shitty and would not fool anyone really. Nothing new, it's a basic chatbot. This should not be news.

See this: http://www.scottaaronson.com/blog/?p=1858
legendary
Activity: 2562
Merit: 1071
The Turing Test should be performed with judges experienced in computer science, artificial intelligence etc, not any random people

That would defeat the point of the test, though the test itself is only somewhat relevant to the whole concept of AI (and badly needs updating).


If a computer is mistaken for a human more than 30% of the time during a series of five minute keyboard conversations it passes the test. No computer has ever achieved this, until now. Eugene managed to convince 33% of the human judges that it was human.

That's not true; this wasn't the first time a computer program was able to fool at least 30% of the people it talked with. For reference: http://www.newscientist.com/article/dn25692-no-skynet-turing-test-success-isnt-all-it-seems.html?full=true

A couple of quotes from the newscientist article:
Quote
But as early as 1991, a bot called PC Therapist created by Joseph Weintraub took part in a Turing test and fooled 5 out of 10 judges into thinking it was human – a pass rate of 50 per cent. Meanwhile, much more recently, in 2011, Rollo Carpenter's Cleverbot chatted with 30 humans in front of a live audience of over 1000 and fooled 59.3 per cent of the judges and audience into thinking it was human.

Quote
"The purpose of this June's Royal Society experiment is not to find if a machine can now pass one interpretation of Turing test success, 30 per cent, especially as a machine surpassed this with a deception rate of 50 per cent back in 1991 and machines do not have anywhere near the conversational skills of HAL from 2001: A Space Odyssey."

Rather, the organisers wanted to "encourage/inspire children to take up computing and robotics, is also about awareness and prevention of cybercrime".
full member
Activity: 144
Merit: 100
What does Passing The Turing Test mean? Indicators of Progress in AI with Robin Hanson!(1:11hrs)
Robin Dale Hanson (born August 28, 1959) is an associate professor of economics at George Mason University and a research associate at the Future of Humanity Institute of Oxford University. He is known as an expert on idea futures, markets and was involved in the creation of the Foresight Exchange and DARPA’s FutureMAP project. He invented market scoring rules like LMSR (Logarithmic Market Scoring Rule) used by prediction markets such as Consensus Point (where Hanson is Chief Scientist), and has conducted research on signaling.
http://www.youtube.com/watch?v=8kzJSDFZAz4
legendary
Activity: 952
Merit: 1005
--Signature Designs-- http://bit.ly/1Pjbx77
Trailer to University of Reading Turing test event, The Royal Society, 7 June 2014.
https://www.youtube.com/watch?v=0Hgw9RVwbaw

This is just a trailer and the demo site is down.
I was searching the net for more info...
http://www.reading.ac.uk/news-and-events/releases/PR583836.aspx

Apparently, only 10 judges out of 30 thought the software was a foreign teenager. I could not find any transcript of the conversations, but this guy found a transcript of the questions and answers for the 2012 test:
http://www.slate.com/articles/technology/bitwise/2014/06/turing_test_reading_university_did_eugene_goostman_finally_make_the_grade.html

Assuming the answers are not that much better in 2014, I think AI conversation is still in its infancy.
full member
Activity: 153
Merit: 100
They lower the expectation and claim the test is pass.


You could lower your kid's score to below average and claim they pass your test also.
full member
Activity: 271
Merit: 101
It didn't pass the turing test.

They were told that some participants didn't speak English as a first language, that some were young and they were given a 5 minute time limit to decide who was a machine.

The turing test doesn't mention any of the above. It should fool 50% of people that it is a real person, not fool 30% of people in a 5 minute time window that it might be a kid who speaks poor English
DrG
legendary
Activity: 2086
Merit: 1035
Just don't ask it to control nuclear weapons when owned by a company called Skynet.
full member
Activity: 144
Merit: 100
AI vs. AI. Two chatbots talking to each other
http://www.youtube.com/watch?v=WnzlbyTZsQY
sr. member
Activity: 406
Merit: 250
Maybe Dank is a computer.
hero member
Activity: 503
Merit: 500
I'm writing a sci-fiction story and I thought about making an AI grow out of some sort of blockchain, because like DNA there is an equal copy of all the data in every computer... Good idea? Any tips? Cheesy
sr. member
Activity: 392
Merit: 250
And what if Internet is intelligent and has gain consciousness, but does not manifest it to us? ;]

What if Satoshi Nakamoto is the internet itself...

Shocked
Could be. Everything is possible.
hero member
Activity: 503
Merit: 500
Don't you guys think that simulating a 13 year old is a way to hide the flaws (but yeah it's an improvement)? Also, we could go philosophical and question the validity of the turing test, it seems a pragmatic yet insuficient way to determine if the AI is intelligent...
b!z
legendary
Activity: 1582
Merit: 1010
The Turing Test should be performed with judges experienced in computer science, artificial intelligence etc, not any random people
full member
Activity: 144
Merit: 100
Trailer to University of Reading Turing test event, The Royal Society, 7 June 2014.
https://www.youtube.com/watch?v=0Hgw9RVwbaw

You can test Eugene Goostman yourself (if you can get through the congestion).
http://www.princetonai.com/bot/bot.jsp
sr. member
Activity: 252
Merit: 250
No it means that Allah the merciful is not satan but life itself  Cry
legendary
Activity: 826
Merit: 1002
amarha
And what if Internet is intelligent and has gain consciousness, but does not manifest it to us? ;]

What if Satoshi Nakamoto is the internet itself...

Shocked
sr. member
Activity: 392
Merit: 250
And what if Internet is intelligent and has gain consciousness, but does not manifest it to us? ;]
full member
Activity: 196
Merit: 101
If a computer is mistaken for a human more than 30% of the time during a series of five minute keyboard conversations it passes the test. No computer has ever achieved this, until now. Eugene managed to convince 33% of the human judges that it was human.
legendary
Activity: 826
Merit: 1002
amarha
So, how does it benefit us?

We will have programs who can act as believable human customer support representatives at some point. Also I'm sure there will be AI that people become friendly with in some way. Personal assistants will become much better than they are now(Siri ect).
full member
Activity: 169
Merit: 100
So, how does it benefit us?
Pages:
Jump to: