When do you think an android will be...

[matt-browne-sfw]

... able to pass the Turing Test and collect the Loebner prize of $ 100,000 ? Or rather the NLP software of an android.

My guess is not before the year 2030. Any thoughts on that?

[Moonbat]

Why 2030, why not later, or sooner.

The last time I saw computers trying to pass the turing test was on Tomorrow's world, and only the internet scientists that quesitoned the 'computers' were sure of which ones were and which weren't. That was less than 5 years ago.
Does turing's test define the questions? becuase if not then the questions will evolve along with the computing. I would be surprise if a robot/software/PC/android would be able to past the Turing test before 2050 at elast, and probably 2100.

But I'm a rarely pessimist so maybe they've already done it but were clever enough to fail the test and keep their obscurity and freedom!

I remeber a story I read once, it think it was from a book that spoke to Sci-fi that has predicted stuff and it was about the internet and computers learning everything and then killing (sigh) and taking over the world, but I can't remeber who it was by. Sorry I went off on a tangent then.

Anyway Matt, no I don't think they will pass by 2030.

Moonbat

[matt-browne-sfw]

Originally Posted by Moonbat

Why 2030, why not later, or sooner.

The last time I saw computers trying to pass the turing test was on Tomorrow's world, and only the internet scientists that quesitoned the 'computers' were sure of which ones were and which weren't. That was less than 5 years ago.
Does turing's test define the questions? becuase if not then the questions will evolve along with the computing. I would be surprise if a robot/software/PC/android would be able to past the Turing test before 2050 at elast, and probably 2100.

But I'm a rarely pessimist so maybe they've already done it but were clever enough to fail the test and keep their obscurity and freedom!

I remeber a story I read once, it think it was from a book that spoke to Sci-fi that has predicted stuff and it was about the internet and computers learning everything and then killing (sigh) and taking over the world, but I can't remeber who it was by. Sorry I went off on a tangent then.

Anyway Matt, no I don't think they will pass by 2030.

Moonbat

Well, Moonbat, it was a guess. An optimistic one, I admit. Looking back in the early sixties there was already talk about chess programs beating humans and also machine translations of natural language texts.

As for the Turing Test itself, the testers can freely choose their questions.

There are also Loebner prizes for best NLP systems of the year and the like. The $ 100,000 can only be collected once.

[Interference]

Any day now.

All right, maybe not next Wednesday, but mimicking the process of reactive thought and conversation is achievable (I, myself, have been called a mechanical beast, but that, I feel, is a story for another day). A computer might already be said to think, though it can not as yet make independently arrived-at choices - like the selection of particular and appropriate words, for example. But that's the basis of conversation. We often use familiar turns of phrase to respond to familiar sounds or experiences. Key words might trigger key responses in an automaton. (Unfortunately, I can remember a sequence in an episode of the Six Million Dollar Man where Steve Austen caught out the robot played by John Saxon because the robot's automatic response came up with an exact phrase and an exact inflection twice in reply).

Making a bi-pedal creature with arms, hands and a head is a question of weight/balance/bald-or-hirsuit. Achievable, why wouldn't it be?

Teaching a robot to type ... hmmm. Tricky.

If that's all it would take to win the prize, I'd say it could happen within the next five years easily if anyone were to put their mind to it. The question is one of desirability. The robots we need aren't an average height of six feet with a limited number of appendages and digits, a paltry pair of eyes and an orifice for sniffles. We need robot arms as limb replacements, robot spanners in construction, robot cockroaches for space exploration, we need something better than an android.

Secondly, who would want an android?

I: Do you know how long it was between the manufacture of rubber products and the first blow-up woman?

II: Can you imagine anything more eery than a mechanical butler that can switch itself off and stand in a corner? Would you, however logically you might know that it was simply a programmable machine, trust one with your children? Or your laundry? Or cooking?

III: Who could/would pay for the luxury of a robot that looks and acts human when you can have a robot that looks like nothing more than a meccano set that you won't feel you need to apologise to if you accidentally kick it?

Robots that seem alive will be unpopular.

But that wasn't your question, the answer to which is really that we already have the technology, but the prize isn't worth it - when we can get one elected President of the U.S. and get it to say and do anything we damned well please (... and you try to warn people, and they just don't believe you ... Look at its eyes! LOOK AT ITS EYES!!!)

[manephelien]

I'd certainly trust machines, human looking or otherwise, with my laundry and cooking. Less certain about kids, however.

They have made a robotic face in Japan which can mimic emotions. It has visual sensors and smiles when you smile at it and frowns if you frown. A far cry from a fully artificial intelligence I grant you, but apparently most people who have had contact with it say it's easier to communicate with it than a standard voice synthesizer. We all come in contact with those when we call automatic answering services, some of which can understand what you say, even if it's just a limited vocabulary.

AI is already used in limited applications and some problem solving (not least computer games) but I doubt they'll be as versatile as genuine human beings for a long time to come. I'm not sure such a development would be particularly desirable either.

Humans will antropomorphize anything from machines to animals (even though paradoxically most animals are far more intelligent than the average person would give credit for, but they are not human).

[matt-browne-sfw]

Someday androids could become crucial for serious interstellar travel attempts. They could raise babies on a starship (who were traveling as frozen embryos for thousands of years potentially).

[matisamd]

it seems simple in planning to me all we need do is understand our selves.

work out a flow diagram that shows how we work ( at first with in boundries that can grow with experiement and understand of our selves compared to the abillity of machine)

and either have the technology to impliment that or the time, as given enough time i bealve its possible to make a steam engine power the universe, type thing thing


theres my 3 rules i am talking about human emulation. If you just want it to drill holes in an alien planet same rule set slightly different criterea less time or tech needed.

My point is we aint so complex that our vanity and ignorance might indicate in our actions and measurable abillities i beleave emulation is as simple as it was to build a nuclear bomb its simple once you know how but seems impossible tell done.