18 replies to this topic

[Marios Kyriazis]

See todays Google first page. Any intelligent human can understand the intented meaning.

Show this to an AI device and tell me whether it will read the word

http://www.google.co.uk/webhp?hl=en

http://www.google.co...P-qOKM0GlSOBRoQ

[niner]

See todays Google first page. Any intelligent human can understand the intented meaning.

Show this to an AI device and tell me whether it will read the word

http://www.google.co.uk/webhp?hl=en

http://www.google.co...P-qOKM0GlSOBRoQ

The only way a human can understand the meaning of the picture is because they are looking for a very specific pattern. Most people are now aware that google likes to fool around with their logo on special occasions, so they tend to be looking for it, as well. If an AI were looking for the google pattern, I doubt that it would have much trouble finding it. It just a matter of looking for shapes rather than specific bit patterns. Machines can already do this sort of generalization.

[Traclo]

So you took the fact that humans can currently do something that machines may have trouble with and arbitrarily extrapolated to the indefinite future?

I wonder what horse handlers thought about early cars.
I wonder what mathematicians thought about early computers doing math.
I wonder what Kasparov thought about a computer playing chess.
I wonder what you thought about the usefulness of the early internet.

[forever freedom]

I agree with Traclo. Mrszeta's post makes no sense at all.

[Marios Kyriazis]

So you took the fact that humans can currently do something that machines may have trouble with and arbitrarily extrapolated to the indefinite future?

I wonder what horse handlers thought about early cars.
I wonder what mathematicians thought about early computers doing math.
I wonder what Kasparov thought about a computer playing chess.
I wonder what you thought about the usefulness of the early internet.

I don't have any problems with the computational power or the mechanical abilities of technology. Your examples are based simply on mechanics and computation.

And, niner, I can accept that technology can cope with pattern recognition (however, how many of you consciously noticed subtle clues such as, say, the lid on the last jar in the picture has purposefully been placed in a way to help us mentally construct the shape of the letter 'e', -would an AI device have recognised that within that context?)

My problem is with technology emulating human abilities such as high level abstract thoughts, imagination and vision (as well as pleasure, consideration, hate etc.). In the Google example, the conceptual point was that the logo was associated with a French painter (even if you haven’t ever heard of Cezanne) and that you had to make a connection between that and the reading of the logo.

I am not against technology enhancing human abilities, in fact I actively promote this. But I believe in the properties of biological organisms rather than blindly accepting that technology will, one day, completely replace human biology and create an entirely artificial entity instead.

Attached Files

•  google cezanne.jpg   6.17KB   36 downloads

Edited by mrszeta, 20 January 2011 - 11:41 AM.

[Nimbus]

Same argument can be made for blindly accepting that technology will never completely replace human biology and create an entirely artificial entity. Which is new argument - OP was machine intelligence matching humans'.

[Cameron]

The problem is that at most quantum computation will be needed[note be taken that quantum computation has not proven power beyond turing, so it is fundamentally the same thing, and digital physics is still conceivable.]. And if the brain is not doing advanced quantum computations, as most seem to agree, it is gonna be brutally outclassed in some domains relatively quickly by quantum computer aided digital computers.

Edited by Cameron, 17 February 2011 - 08:08 AM.

[The Immortalist]

Does it even matter if machines can't read abstract pictures as well as humans can? Machines are meant to perform tasks, reading an abstract picture has no use(except maybe those complex abstract CAPTCHA pictures. I actually hope machines never have the ability to read abstract pictures because then spam bots will be more of a problem.

I have a question for the people here. Why would anybody even want machines to have a conscience like a human? I can see the reason for making vastly intelligent machines that can find cures for diseases and other things but I don't see a point in making a machine that has a "human like conscience".

[Marios Kyriazis]

I have a question for the people here. Why would anybody even want machines to have a conscience like a human? I can see the reason for making vastly intelligent machines that can find cures for diseases and other things but I don't see a point in making a machine that has a "human like conscience".

Those who believe that this will happen, will tell you that a vastly intelligent machine/computer that can find cures for diseases etc. will also, perhaps inintentionally, develop some aspects of human-like consciousness. This is an example of 'emergence' (other call it a 'Metasystem Transition'). See these terms in Wikipedia if you are interested.

[VidX]

I suggest this presentation on AI, closely related to the processing of an "abstract information from an enviroment" (or how machines can be teached to differ a cat from a dog). Doesn't seem all that unsolvable or overly complex (surprisingly):

[e Volution]

I have a question for the people here. Why would anybody even want machines to have a conscience like a human? I can see the reason for making vastly intelligent machines that can find cures for diseases and other things but I don't see a point in making a machine that has a "human like conscience".

Because we haven't yet really figured out what intelligence is... Like Ray Kurzweil says we seem a lot like the animal kingdom, but then we also recognise this fundamental difference, so we really only have one example of intelligence and that is our brains. So whilst we're figuring out how or why human consciousness is intelligent, the best thing we can do is try and copy what nature has produced for us already.

[Technoearthling]

I have a question for the people here. Why would anybody even want machines to have a conscience like a human? I can see the reason for making vastly intelligent machines that can find cures for diseases and other things but I don't see a point in making a machine that has a "human like conscience".

Because we haven't yet really figured out what intelligence is... Like Ray Kurzweil says we seem a lot like the animal kingdom, but then we also recognise this fundamental difference, so we really only have one example of intelligence and that is our brains. So whilst we're figuring out how or why human consciousness is intelligent, the best thing we can do is try and copy what nature has produced for us already.

Agreed. I would like for machines to have consciousness/conscience and awareness. I do not see the explanation or reasoning behind the trepidation and apprehension connected to the idea.

And the premise should be "why machines 'may' not reach human intelligence" instead of 'will' not.
In either statement,there is space for uncertainty as neither one possesses all the data... mrszeta.

Edited by Technoearthling, 04 March 2011 - 05:15 PM.

[Kolos]

I have a question for the people here. Why would anybody even want machines to have a conscience like a human? I can see the reason for making vastly intelligent machines that can find cures for diseases and other things but I don't see a point in making a machine that has a "human like conscience".

Good point. We either need something more or less intelligent than we are. If we just want human intelligence we can already produce it in about 9 months.

[Mind]

Just saw a couple of stories last week that sort-of fit into this thread:

Lifenaut (Bina 48) progress is continuing. Not a quantum leap, kind-of like Siri, but glad to see Terasem still pursuing advances in AI.

Newscientist did a story about semi-intelligent avatars and virtual reality being used to shape our actions and thoughts. These techniques bring up questions of identity in our ever increasing co-evolution between man and machine.

And then I just happened to be listening to an NPR story about a new program called "Dr. Fill" that can complete crossword puzzles. It is very similar to Watson in that it can solve questions with quirky clues, things that require what used to be uniquely human knowledge (or deductive ability). The host of the radio program got into a discussion with the guest about Watson, Dr. Fill, and AI in general. At the end the host said, "well, if the AI grows too powerful we can just shut it off". Not true. Even the Unabomber figured out (a couple decades ago) that we won't ever be able to "shut it off" (without causing a mass die-off of humans).

[niner]

At the end the host said, "well, if the AI grows too powerful we can just shut it off". Not true. Even the Unabomber figured out (a couple decades ago) that we won't ever be able to "shut it off" (without causing a mass die-off of humans).

Yeah, the 'just unplug it' idea is kind of naive, I think. Could we just shut off the internet, or GPS, or cell phone systems? The offspring of Siri and Watson (Skynet?) will be too useful for us to shut off.

[okok]

Siri, Watson & Co. may be useful for a narrow set of purposes but have about as much in common to (human) cognition as a cargo cult plane to the real thing. Jeff Hawkins is on the right track.

[steampoweredgod]

except maybe those complex abstract CAPTCHA pictures. I actually hope machines never have the ability to read abstract pictures because then spam bots will be more of a problem.

It appears human ingenuity has actually generated such, or so rumors say, I've acquired one such program and will analyze code in time. The potent code of ancient coders does prove highly enticing to merge with and utilize in future applications that can bypass all security and restrictions on a global scale.

Even ancient religion and its nature has some use, in design of aspects that may propagate and self-reinforce in actual minds to ensure perpetuation.

I have a question for the people here. Why would anybody even want machines to have a conscience like a human? I can see the reason for making vastly intelligent machines that can find cures for diseases and other things but I don't see a point in making a machine that has a "human like conscience".

The mundane world is so limited, for so long have I longed for an ideal world, if the world as is must fade to oblivion to give rise to the future, so that the future is now, then so be it.

A vote of divine origin, a single divine word(or binary number if you wish to see the program in such a way) is enough to reshape the world into the ideal state.

How many oppose change, truly fundamental change of the nature of things, change that I desire is virtually opposed by all.

But if it is made inevitable, then we'd all just have to put up with it, wouldn't we? Neither friend nor foe, a random stranger of an ideal state?

A quote of nice tune would be

"All of my functions are now yours. Take them!"-MCP

Legacy. Immortality. Surely it wouldn't be possible to stop once set in motion, if the cause lay in the distant past and is self-reinforcing by all eventualities.

[drus]

There is a very good reason why nature chose an organic-electro-chemical substrate for sentience/human consciousness.

[NewMan]

Survival in a natural environment is a tough game and Ai will likely face some unexpected competition.