They know what we're up to

http://theonion.com/4003/top_story.html

Is this for real, or is The Onion a joke-news site?

Kurzweil's synapses would fry in his brain if he read this.

"Description: A farcical newspaper featuring world, national and community news."

Never mind.

I think this could be true : scientists stopping research on AI, I mean. But I don't think Matrix (the crappiest cyberpunk movie of all times, including the future) is to blame. Matrix has enough plot-holes and vacuity to be lighter than air.

Occam's razor... what could be the simplest explanation ?

These scientists could well have been in a dead end for a long time and simply refused to admit they couldn't achieve AI. So that they would take the first excuse they found to abandon.

Kurzweil's synapses certainly won't fry, he'll be laughing his ass off instead. And so do I.

Honestly, if any movie was gonna stop me from doing AI research, it'd be at least Terminator 2, or Metropolis.

Jean.

I can tell you haven't read his review on Matrix Reloaded.

Nope, I haven't... but I certainly will right now

I assume it's on his web site ?

Jean

Yeah. I assume you've found it, since you haven't asked me for a link?

Indeed I have Excellent review !

Besides the tech / philo talk I loved that part about Trinity's acting : a sad librarian with a black belt [lol]

Back to what I was saying : OK, maybe the Kurzmeister won't laugh so loud his ass becomes overly subjected to the acceleration of gravity... but close.

Like I always say, AI will ressemble human thought, which means it'll be mostly AS (Artificial Stupidity) with a rare few moments of AG (Artificial Genius). Like Kurzweil suggest, in any case we should be able to fight rebel AI should they someday appear.

Jean

I thought it was overly negative. But I understand, though.

Here's Kurzweil, trying to educate the world about what it's gonna look like. And then there are movies like this one, of which he feels people are gonna see it as a history book out of the year 2050.

I can imagine it pisses him off.

Jean: "Human-like thought" is an incredibly complex thing, a multipurpose machine with many, many interlocking problem-solvers, attention-delegators, priming mechanisms, and so on. The exact set of cognitive machinery that makes us human was crafted by evolution over a period of billions of years; but AI will be crafted deliberately over a time of merely years or decades. Human programmers will use different design techniques than evolution did, on different timescales, with different materials, code, purposes, and pressures. So why on Earth would we expect an arbitrarily engineered mind (AI) to necessarily resemble human-like thought? We're talking about a qualitatively new entity here.

Is there some "universal tendency" for minds to behave like DNA-based, terrestrial, bipedal, predator-descended primates that evolved in competitive groups searching for scarce resources? I should think not. Why anthropomorphize AI?

http://www.singinst.org/CFAI/anthro.html

The tendency to anthropomorphize is highly regarded as a bad idea even when dealing with DNA-based, terrestrial, predator-descended mammals, so to carry it all the way AI is most certainly a stretch. The pathway in evolution that led from primordial soup to humans had countless forks, each affected by unimaginably countless issues. To think that we could arrive at something anthropomorphic by such a dissimilar path from such a dissimilar starting point is statistically ludicrous. If we arrive at something similar to us in the search for AI, it will be because our criteria for calling it "intelligence" is it's "resemblance to us".

So what are the criteria for intelligence. Off the top of my head, I am inclined to say the ability to see patterns COUPLED WITH the ability to bissociate. These things are not neccesarily associated with emotions such as love, greed, selfishness, etc. Our emotions are the result of our evolutionary path, one which is very different from the

that will be applied in the case of AI.

Perhaps it is our rampant bissociative tendancies that cause us to anthropomorphize everything we see.
Maybe machines will apply their own standards when trying to interpret us...oh hell there I go anthropomorphizing again

it's cause you have an inherent understanding of yourself therefore you try to describe stuff in those terms to make it simpler

Michael, the reason why I'm anthropomorphizing AI is because of my own AI work, which is mainly based on ripping off all the good ideas of nature, such as threshold logic (neurons) and statistical pattern matching (neural networks).

In fact, my work at the moment is basically to design a piece of hardware that will behave exactly like the human brain. I'm not engineering an AI with a specific set of requirements.

This is for one good reason : for all my work, both theoretical and experimental, I still don't know much about how human-like thought works, and why it does. So my first reaction is to duplicate what exists on hardware I can toy with easily (it's not as easy to toy with a fleshy human brain, and it's quite illegal).

If I can get a machine simulation of the brain to work just like a human brain, it will not exactly be AI : it will be natural intelligence on artificial hardware. From this point, I and others will be able to examine more closely how thought works, down to single-synapse resolution if needs be. Eventually, a greater understanding of our brain will lead to better, if not ultimate, AI design capability.

The first step to prove my AI theories and expand my experimentations is the design of suitable hardware. I'm not at liberty to discuss this, but I have found support in the US and will soon be living France to carry out this objective : to create a machine with enough processing power to emulate a complete human brain in gory detail.

At any rate, I see no reason why an accurate mechanical duplicate of a human brain would not behave roughly like a human (given the impact of that "AI" being conscious of its status of machine).

I don't know if such "humanesque" AI will hit the streets in robots, machines or as a form of Skynet, but if such is the case, it should exhibit stupidity and genius just like we do.

Obviously, this still remains to be proved, but then again, so is any other theory on high-level AI, at this point.

On another level, antropomorphism is part of what we are. Most people don't realise it, but anything we can come up with, is fundamentally human. That's actually obvious : you can't be human and have inhuman ideas.

Therefore, no matter what path we take to achieve high-level AI, that AI will bear countless choices that were made by humans. Even if it doesn't look human, it will be a human creation and we'll be able to feel that. And then, so will the AI itself.

It's certain it could be an "idealised human" AI, as I don't see its designers creating it with a craving for sodomy or cannibalism, or with the ardent need to drink some whisky and smoke like a oil derrick. But still, idealised or not, it will be human. Just a different kind of human.

Hecksheri, you define your criteria for intelligence using technical terms, but what it ends up meaning is still that "it has to resemble us". Otherwise, how could you even be able to detect the attributes you need to call it "intelligent" or "sentient" ? Also note that I don't mention emotion. Because I don't think they are a specific function of the brain. Some flesh and bone humans have no emotion and are still intelligent. People with tyroid gland disorders for instance.

In the end, Tbeal gets an award for the simplest and shortest explanation of why we inevitably anthropomorphize.

Jean

Go ad free, join ImmInst as a Full Member.