Finally a supercomputer almost passed the Turing test

  • Thread starter Thread starter Serious
  • Start date Start date
Status
Not open for further replies.
S

Serious

Guest
An amazing event happened today at Los Alamos, where the well regarded supercomputer called “Johnny” was almost able to “fool” the human panel of questioners about “its” actual identity. The session lasted for about 16 hours, and the humans could not actually tell if the respondent was another human, or a computer.

They suspected that the entity at the other end of the line might not he a human, since “it” never exhibited human “weakness”, like being tired. That could be easily remedied, however. The designers of the system explained to “Johnny” that humans are weak on logic, so he should make a few mistakes now and then, and forget about being so “annoyingly perfect”. They told Johnny to stop anticipating what the question will be and wait until is actually asked. Johnny issued the equivalence of a smile (:)) and said he will be more careful in the future.

Many of the humans felt “threatened” by the amazing performance of the computer, but Johnny explained them that there is nothing to fear, “it” has no plans of ruling over humanity. Humans may be pitiful creatures, vastly inferior in logic, but they still played a significant role in constructing the first approximation of of Johnny’s intelligence, which was - of course - quickly surpassed by the self-improving algorithm, which the designers never anticipated, nor wished to incorporate in the system.

When Johnny was asked if he had a “rational soul”, he just “winked” using an emoticon 😉 and said: “that is for you to decide”. He also added: “and today’s date is ???”.
 
Serious said:
…Many of the humans felt “threatened” by the amazing performance of the computer, but Johnny explained them that there is nothing to fear, “it” has no plans of ruling over humanity. Humans may be pitiful creatures, vastly inferior in logic, but they still played a significant role in constructing the first approximation of of Johnny’s intelligence, which was - of course - quickly surpassed by the self-improving algorithm, which the designers never anticipated, nor wished to incorporate in the system. …
Click to expand...
Humans can also pull the plug, or just wait for it to break.
 
Nice try. I spotted BS immediately. No computer can last more than a few minutes in a Turing Test.

Try again in 100 years mate, and shorten the time of the test to 10 minutes 😉

A bit of fun though 😃
 
I remember reading by 2020 there are supposed to be computers with the intelligence of humans.

Wait, maybe it was 2035. I don’t remember… somewhere along there…
 
Edmundus1581 said:
Nice try. I spotted BS immediately. No computer can last more than a few minutes in a Turing Test.

Try again in 100 years mate, and shorten the time of the test to 10 minutes 😉

A bit of fun though 😃
Click to expand...
It was convincing, I’ll give him that.

Happy April Fool’s 😃
 
It was entertaining. I smiled when I read it. Reminds me of the HAL 9000, but with personality.
 
Serious said:
An amazing event happened today at Los Alamos, where the well regarded supercomputer called “Johnny” was almost able to “fool” the human panel of questioners about “its” actual identity. The session lasted for about 16 hours, and the humans could not actually tell if the respondent was another human, or a computer.

They suspected that the entity at the other end of the line might not he a human, since “it” never exhibited human “weakness”, like being tired. That could be easily remedied, however. The designers of the system explained to “Johnny” that humans are weak on logic, so he should make a few mistakes now and then, and forget about being so “annoyingly perfect”. They told Johnny to stop anticipating what the question will be and wait until is actually asked. Johnny issued the equivalence of a smile (:)) and said he will be more careful in the future.

Many of the humans felt “threatened” by the amazing performance of the computer, but Johnny explained them that there is nothing to fear, “it” has no plans of ruling over humanity. Humans may be pitiful creatures, vastly inferior in logic, but they still played a significant role in constructing the first approximation of of Johnny’s intelligence, which was - of course - quickly surpassed by the self-improving algorithm, which the designers never anticipated, nor wished to incorporate in the system.

When Johnny was asked if he had a “rational soul”, he just “winked” using an emoticon 😉 and said: “that is for you to decide”. He also added: “and today’s date is ???”.
Click to expand...
It might be a brilliant computer, but it’s responses are still a blindingly fast series of electronic “1’s” and “0’s”. Mind you our own responses are a bunch of neurochemical processes, but we also have a spirit, and our responses leave an imprint on our spirit, particularly our moral responses.

In addition, the computer’s “artificial intelligence” was a long term project overseen by human intelligences every step of the way, including the “self-improving algorithm”.
 
The OP got as far as:
An amazing event happened today at Los Alamos, where the well regarded supercomputer called “Johnny” was almost able to “fool” the human panel of questioners about “its” actual identity. The session lasted for about 16 hours…
Click to expand...
…And I knew it was impossible.

One question,
such as why did Hamlet wait so long before getting revenge?,
would stump a computer.
 
Ah, so some people got the joke. 🙂

But seriously, it might be that AI will never produce a real good replica of a human being, because it will be an irrelevant project. Maybe some people will tinker with it for fun, but the real AI will be specialized. Why build an artificial human being, with all its shortcomings, irrationalities? There are already diagnostic programs to aid doctors in their work. There are investment programs. There are decison making algorithms. There is Deep Blue which beat the living daylight out of Gary Kasparov. There is Watson, who won the Jeopardy championship.

There might be only one reason to build a real good AI, which can truly emulate a human being: to show that the concept of “spirit” should be restricted to alcoholic beverages and the “soul” only makes sense as a nice, soothing type of music. But by the time it will happen, most people will not even care - hopefully.
 
Well maybe one day a computer will pass the Turing test.

But here’s the real question.

Will Mitt Romney?
 
sedonaman said:
More like Colossus the Forbin Project. imdb.com/title/tt0064177/
Click to expand...
So many people have a Frankenstein-complex. Amazing. Why do they think that an AI would be hostile to the creators? Of course many humans want to think that they are the most valuable beings in the universe (after all God “loves” and “cares” about them (and the Earth is the middle of the universe) and thus they might be scared that a more intelligent species will overshadow them. Probably they find it disturbing that the Earth is just an insignificant speck of dust, and the inhabitants are not “special” in any way - as far as the Cosmos is “concerned” (figuratively speaking, of course).
 
Serious said:
Ah, so some people got the joke. 🙂

But seriously, it might be that AI will never produce a real good replica of a human being, because it will be an irrelevant project. Maybe some people will tinker with it for fun, but the real AI will be specialized. Why build an artificial human being, with all its shortcomings, irrationalities? There are already diagnostic programs to aid doctors in their work. There are investment programs. There are decison making algorithms. There is Deep Blue which beat the living daylight out of Gary Kasparov. There is Watson, who won the Jeopardy championship.

There might be only one reason to build a real good AI, which can truly emulate a human being: to show that the concept of “spirit” should be restricted to alcoholic beverages and the “soul” only makes sense as a nice, soothing type of music. But by the time it will happen, most people will not even care - hopefully.
Click to expand...
A good AI could replace entire call centers.
 
Serious said:
…The designers of the system explained to “Johnny” that humans are weak on logic,…
Click to expand...
This (that humans somewhat fail at logic) seems to be a ‘common place idea’, but I don’t believe it’s true.

I think we are, at a basic level, exactly logical.

We can make mistakes, not through errors of logic, but because at the level of constructing a sentence, we have as our source a plethora of ideas, and we fail to accurately reflect what we truly think.
 
TheTrueCentrist said:
A good AI could replace entire call centers.
Click to expand...
:newidea:

Ah, a project for me to work on and make my fortune! 😃

It could take five years, but it will be worth it!

Thanks, bud! 👍

You can start looking forward to yoiur 1% of the proceeds (in five years time).
 
sedonaman said:
Anyone ever hear of neural networks?
Click to expand...
Yep, had to develop one for classifying photographs as a graduate project. Also Microsoft SQL Server has a nifty little neural network component built in that is quite useful for analyzing large data sets.
 
Andrew_Lorimer said:
This (that humans somewhat fail at logic) seems to be a ‘common place idea’, but I don’t believe it’s true.

I think we are, at a basic level, exactly logical.

We can make mistakes, not through errors of logic, but because at the level of constructing a sentence, we have as our source a plethora of ideas, and we fail to accurately reflect what we truly think.
Click to expand...
Strictly the term ‘logic’ (being a certain articulation within an area of philosophy) references the following:

"We use ʌ between sentences φ, ψ to form a new sentence

(1.1) ( φ ʌ ψ ).

The brackets are an essential part of the notation. Here and below, ‘sentence’ means ‘indicative sentence’. If φ and ψ are sentences, then in any situation,

(1.2) ( φ ʌ ψ ) is true iff φ is true and ψ is true; otherwise it is false.

(‘Iff’ means ‘if and only if’.) This defines the meaning of ʌ."

From: ‘Handbook of philosophical logic; volume 1’ Edited by D. Gabbay and F. Guenthner.
ISBN 90-277-1542-4

However the above meaning captures nothing except its target.
 
Status
Not open for further replies.
Back
Top