Also, world-renowned biologist PZ Myers and cognitive scientist / computer genius John McCarthy ("father of AI") are both extremely critical of Kurzweil.

Seems like most of the cognitive scientists and philosophers of mind are very critical, or even completely dismissive. His followers, in a sense, seem to be mostly computer scientists. This is not a coincidence obviously.

So your point that he can't be wrong is that he's an incredibly smart dude (who might be extremely wrong)? What is your point in this thread again?

For all you guys psyched about the similarity read "I Have No Mouth, And I Must Scream" then tell me how awesome it's gonna be to live forever controlled by computers.

http://en.wikipedia.org/wiki/I_Have_...st_Scream#Plot

SURVIVORSHIP BIAS

Wow people...come on.

It's the same thing with financial advisors picking mutual funds. Take 1000 futurists and statistically speaking, just from them guessing, we'll expect a few of them to show remarkable success. Then, it will naturally seem as though they're really really talented but it was just luck.

This may not necessarily be so big of a question. Humans will adapt, and if we get annoyed by the evolving robots/computers, we will just legislate against them, forbid developing of more effective robots/computers. That´s what I meant with "pulling the plug" earlier in this thread. I think though, that humans benefit from and like the technology that much that it will sneak in by time in the lives of humans anyway. Robots taking care of the sick and so on. The movement against the robots will give time to adapt to and refine the concept. We will not get a Singularity in the sense Kurzweil predicts.

You realize that the nature of this message board is such that there will be a large number of people who are predisposed to not like to think about what survivorship bias means.

Yeah...it's so hard to distinguish between skill in an area where chance plays a huge role vs survivorship bias. It's also related to a comment of mine in another thread about how we're uncomfortable owning up to the large role that chance plays in much of our successes.

I think some of Kurzeil's ideas are valid, but he goes over the top. I see no reason why the current pace of technological development should continue. “There is no reason why we should not run out of things to discover.” In fact I think the rate of technological advancments will eventually ground to a relative halt.

However certain things that we can see a path to should be possible, like immensity increased life spans.

Further his predictions are falsifiable. Just wait fifty years and look back, should not be beyond most of the population

Cliffs notes: argue whether Moore's law is valid and if strong AI is possible, not the eschatological stuff
Long version:

Kurzweil's criticisms are of a completely inappropriate magnitude.
Here's an example:
There are several comments in this thread that imply that Kurzweil's predictions are as bad as Christian / New Age eschatology.
These criticisms are terrible.

Here is an example:

This is a hypothetical ad hominem attack based on the guilt by association fallacy.
You're saying "Kurzweil looks like all the other 'doom sayers,' and false prophets readjust their dates to the future when they're proven wrong, so Kurzweil will do that too."
There are at least a dozen similar posts in this thread, including those which suggest Kurzweil is a "delusional liar," or that his claims belong at the "bottom of the 2012 slagheap."

Very little of this thread actually addresses the weak and obviously contestable points in his argument.
His argument is:

Step 1) Exponential growth will continue, and reach a point x in the future
Step 2) At point x, we can simulate brains
Step 3) Prophet

Durkadurka is one of the only critics to have addressed step 1 - he says that this is "a fallacy of extrapolation." Unfortunately, his argument ends there. (Hardball47 also makes this point, Princeofpokerstars says something similar).
He does not provide an explanation for why Moore's law has been a good predictor for decades after is formation, nor does he provide any theoretical basis for why Moore's law will end (some would argue "has ended").
There is no explanation for why Kurzweil can't extrapolate here.

Step 2 has been argued a few times in this thread, but again it is a minority of the criticisms.
Durkadurka brings this up:

This is a very brave claim (that the mind can do things machines can't).
I am pretty sure it's a minority belief in neuroscience right now. Dualism is dead.
One poster brought up Searle, but there isn't an argument in this thread for why strong AI isn't possible. In fact, from what I can tell, only Durkadurka seems to explicitly stated that it might be impossible.

Step 3 is where the majority of criticism is directed. The most interesting criticism related to this step is the "I Have No Mouth But I Must Scream" style of argument - namely, that the outcomes of the singularity are gonna suck. It is however, a completely invalid argument (it's a strawman). Kurzweil's usage of the word singularity is meant to capture the unpredictable nature of the event (light doesn't escape a black hole, so you can't find out what's inside). His very argument is that we will not be able to predict the changes, just that the changes will be near-inconceivable. He does not predict a utopian transhumanist future - he predicts a very unpredictable future.

The distribution of argument doesn't make any sense from a logic point of view, and it is certainly inconsistent with how we examine other predictions of similar magnitude.
Predictions on such a macroscopic scope are not always cult-inspired garbage.
Global climate change alarmists, for instance, make extrapolations and present a mechanism to make a massive claim, ranging from "the arable land on Earth will change location, causing global warfare resulting from population movements following the resource reallocations" to "the planet will be flooded and engulfed in an endless series of super-disasters, along with global warfare and terrifying outbreaks of disease, famine, etc."
Although climate change is obviously a topic for another thread, I'd like to point out that the debate tends to a far more civil discourse over the assumptions of the mechanisms for climate change, and the extrapolations of relevant variables (CO2 emissions, etc).

Climate change is far from being the only monster-sized prediction. In political science, we see a spectrum of plausible scenarios ranging from Fukiyama's "The End of History," to Freidman's "The World is Flat." You won't see any legitimate intellectual dismiss Fukiyama's theory on its parallels to silly predictions, despite the book's shocking title.

I suspect this disconnect is due to Mitchell Kapor's quote on the singularity:

"It's intelligent design for the IQ 140 people. This proposition that we're heading to this point at which everything is going to be just unimaginably different - it's fundamentally, in my view, driven by a religious impulse. And all of the frantic arm-waving can't obscure that fact for me, no matter what numbers he marshals in favor of it. He's very good at having a lot of curves that point up to the right."

This is an elegant and poignant piece of rhetoric, and I think it's responsible for a great deal of the anti-Kurzweil sentiment in the (armchair?) intellectual community. Kapor's criticism amounts to a position taken by a smart and successful man with no exceptional scientific or philosophical ability. His qualifications are roughly equivalent to Kurzweil's (elite undergraduate degree, success in industry).

Who can argue with someone who is willing to criticize the intelligence of people with an IQ over 140? Gee, he must have an IQ waaaay higher than that!

if you're interested in contrasting views, you should check out radical evolution

That was a totally disingenuous analysis of my critcisms.

I haven't made an ad hominem of associating the structure of his argument to those of creationists. I was responding to another poster's comment by way of a humorous analogy (to which there are important disanalogies between it and Kurzweil's argument). Furthermore, Kurzweil's prediction really DOES have the structure of being unfalsifiable since it's open to him to merely keep moving the "due date" forward indefinitely by merely making minor tweaks to his theory. This is totally different from how you've represented my criticism.

I don't have to explain how Moore's Law has been heretofore successful in making predictions. It's easy to enumerate cases in the history of science where similar "laws" were proposed that fit the data and then suddenly stopped working! He noticed a trend, proposed a function that describes it (though not an explanation) and hypothesized that it will continue in the future. That's how science often works. But, we have very good reason to not put much faith in such arguments (give the track record of similar predictions).

To this people respond with his track record...to which I respond with: survivorship bias.

Looks like an interesting book. We have to remember though, that whatever is written, things will develop their way. Even if Zeno locked this thread, I wouldn´t care much. It´s just a question of time.

Really? A brave claim?

Seems obvious to me and the reason why so many people dismiss the singularity claims as silly.

FWIW, I do not dispute the possibility that machine intelligences deserving the label 'sentient' will exist some day. I just think it is very far from inevitable and no graphs of increases in processing power will change my view on that.

Yes, I think so.
The claim that the mind can do things machines can't is borderline dualist.
What is so special about a flesh and blood neural net that it's theoretically impossible to simulate it on a computer?

I agree with your wiew on this.

Why is it dualist? A computer is a machine. The brain is a machine, but it may not be a computer.

Machine =/= computer. A computer is a very special kind of machine: one defined by the Church-Turing thesis. So, while the brain may be able to do all that a computer can (if they had enough time and paper, etc.) it doesn't follow that a computer can do all that a brain can do. You can be a functionalist about mind and still deny computational functionalism. You don't need to be a dualist to do that.

I'm sure that we can build a thinking machine...but it won't be merely a computer. You seem to have conflated computer and machine.

I don't think it is theoretically impossible. I just think that it is not inevitable.

This has nothing to do with dualism (at least, for me).

We do not understand consciousness, and if you are talking about machine intelligence then you are talking about consciousness. Since we do not know why we are conscious or how we are conscious, we cannot speak with certainty about the likelihood of machines becoming conscious.

This has no impact on whether or not machines can become conscious; it just affects the degree of certainty we can have when making predictions.

I'm specifically talking about simulating a physical system (neurons).
Suppose we mathematically simulate neurons, glia, etc.
What's the difference between the simulation and a real brain? If you are a functionalist, you probably think that a proper simulation is conscious.

I don't see why a computer can't accomplish this, but, as you imply, a machine could.

For that matter, I have difficulty imaging something an analog computing device can do that a computer can't with sophisticated software / hardware.

So yes, I'd appreciate an explanation for why there's something a machine could do that a computer can't.

You don't understand Searle's criticism of strong AI. It's not analogue vs digital. It's about computing vs thinking. His criticism is that thinking is more than just computation. I don't think that you understand this.

This is what I meant when I said that maybe the brain can do things that a computer qua computer can't (ie, maybe the brain can do non Turing computable functions).

I agree completely.

Whether we can easily simulate a brain is extremely speculative. There are plenty of potential avenues to try to use bio-mimicry to generate the simulation, but whether or not imaging / development biology will be "caught up" enough to provide this data is another matter. I don't know if we can expect the connectome to be finished by 2050, although it seems that it will eventually be finished (unless civilization is screwed).

I'd point out though, that saying "I don't think we'll have a good enough understanding of the brain to cheaply simulate it" is far different from saying "he's a delusional eschatologist liar who has views as laughable as the FSM."

If you can simulate neurons 'in silico,' and use this to make a simulated brain, and a brain is all that is required to think, then can we simulate thinking?

I am familiar with Searle's argument, but it has very little to do with what I asked you.

Clearly, I think Searle is wrong.
I think that the "brain simulator reply" is a valid counterexample. Obviously you do not.
Lastly, I think that you need a magical property in the brain to justify Searle's argument (or any dualist argument). Otherwise, Searle's argument seems to indicate that human brains don't understand anything (just replace Chinese room with "some Chinese dude's neurons and synapses"), and unless you're a dualist, you probably think you are your brain.

"Brain simulator" isn't a counterexample since you need to beg the question that that WOULD be simulating the brain. What if it's impossible? Then it's not a counterexample. Your "counterexample" depends on being able to do it...and whether it can be done is the very question we're interested in!

You don't understand Searle's argument. He's not committed to being a dualist.

This is not a good argument. It's not like mutual funds where you have millions of scam emails increasing the sample size of "guesses" to ridiculous proportions.

There are only so many people in the world well versed in information technologies. Out of those, only so many are actually qualified and intelligent enough to make real predictions, and out of those, only so many do make predictions.

Kurzweil is one of few extremely qualified, extremely intelligent people, and he's been at the forefront of this game for many, many rounds of guesses. He has proven himself as a brilliant inventor decades ago, has earned tons of awards, and has earned the respect of some of the smartest, most qualified people in the world, ranging from Bill Gates to Bill Clinton.

People make arguments about not being able to extrapolate Moore's Law, but if you read the Singularity is near, there is a ridiculous amount of discussion about this. I'm not well versed in this area, so forgive me in advance, but he goes into detail discussing all the possible improvements we can have to our current "flat" computer chips, as there is indeed a limited amount of improvement in those specifically. He talks about more 3d designs, the possible different materials we can use, and even the potential of quantum computing.

The truth is that Kurzweil makes predictions that span a wide array of fields, and the experts in those fields generally agree with him (or possibly fed him the predictions in the first place if you want to be cynical).

If you were to research Aubrey De Gray and read his books, you would understand that Kurzweil isn't just guessing, and that all his data is verifiable.

The people that disagree with Kurzweil are guys like Mitch Kapor of Lotus Software (have you heard of it? I hadn't), and a whole bunch of authors.

The guys that agree with Kurzweil are world leaders in their fields, like the aforementioned Bill Gates.

Maybe I'm bias because of all the TED videos I watch, but I can't tell you how many times I've watched a TED video and gone, "Wow, Kurzweil predicted this back in the mid 90's, and now we basically have proof that it's currently possible in early stages, and will be on the market within a decade".


Honestly, the stuff that impresses me the most in all this, are not the predictions for the future, but the insane **** that we are already capable of today that just hasn't been perfected yet. We've had over a hundred cars travel cross country without a human driver, and this is just something people still say is impossible, despite having been done over and over already.


And on the most basic level of all, just look through history. You've all heard the argument. 200,000 years to discover fire. 50,000 years to discover the wheel. 10,000 years to learn to create metal.... and more recently that has turned into 40 years to spread the telephone... 20 years to spread the microwave... 6 years to spread personal computers... 2 years to spread cell-phone application/personal computing.

How can you be aware of the above and NOT think some super ridiculous insane **** is going to happen in the next 30 years? People 100 years ago would not have believed today's world to be possible in such a short period of time. I feel the same thing is even more true for the next 30 years.

Well, Searle wouldn't have accused me of begging the question.

He takes the Brain Simulator Reply seriously, and does not challenge whether a brain can be simulated.
He (poorly) responds to the brain simulator argument in "Minds, Brains, and Programs."

http://docs.google.com/viewer?a=v&q=...f4-K4fCwV2qcRg

Notice how he starts his argument by stating that this solution (brain simulator) for AI clearly isn't the aim of Strong AIers (who want to achieve AI without understanding neuroscience). He proceeds to generate his own version of a neuron-simulator, but uses pipes and valves instead of something digital. He then says that this giant network of pipes obviously doesn't understand anything, and proceeds to the next argument.

My objection is at this point - I don't think that the comical appearance of the brain simulator prevents it from being conscious. His example is a bit of a strawman - he makes the simulation of the brain rigid and cumbersome, while a real simulation of a brain would allow for growth, adjustment, etc. A real brain simulation would be capable of remembering (and forgetting), learning, emotions, etc.

If he had a proper brain simulation, you could do more than chat with it in Chinese - you could ask how it's feeling, have it write poetry, and watch it become angry at Searle's denial of its ability to understand.

His general argument is that all computers can be realized by "obviously" non-intentional things (Chinese rooms, toilet paper + stones, etc).

He sums his argument as "The problem with the brain simulator is that it is simulating the wrong things about the brain. As long as it simulates only the formal structure of the sequence of neuron firings at the synapses, it won't have simulated what matters about the brain, namely its causal properties, its ability to produce intentional states."

In my opinion, Searle is wrong. I think his characterization of the simulated brain is naive - he refers to it as a "sequence of firings," when clearly a simulated brain is a bit more than that.

Searle states that the physical / chemical existence of the brain is significant for intelligence. Because Strong AIers are willing to use any substrate for their simulation, he claims this makes them 'strong dualists,' and claims this is funny because AIers accuse their opponents of being dualists.

To this I'd reply: Searle, you think that there is something special about the brain as a substrate for intelligence. AIers think anything can be a substrate for intelligence (he explicitly states that this is and says it's a mistake). If something can behave in a fashion which is indistinguishable from the physical brain, who cares if it's made of Play-Doh or transistors? To care seems to ascribe a magical property to the brain that doesn't existence in reality (can not be measured), hence I think you're a dualist.

So, I say "nuh uh! You're the dualist!"