What's on your mind?
General interest discussions, not necessarily related to depletion.
by ian807 » Sun 02 Jan 2011, 19:02:12
$this->bbcode_second_pass_quote('EnergyUnlimited', '')$this->bbcode_second_pass_quote('Article posted by Carlhole', 'H')owever, if we take the possibility of successful human brain emulations this century
seriously, and find some plausibility to the above discussion, more rigorous and extensive
analysis would be well worth the effort.
That is perhaps the most important bit.
We are dealing with academic speculation and not with credible ongoing project.
It is just theoretical
what if... discussion.
Um... No.
http://bluebrain.epfl.ch/
-
ian807
- Tar Sands
-
- Posts: 899
- Joined: Mon 03 Nov 2008, 04:00:00
-
by ian807 » Sun 02 Jan 2011, 19:09:17
$this->bbcode_second_pass_quote('Arthur75', '')$this->bbcode_second_pass_quote('Dezakin', '
')
You aren't even wrong, you're not making any argument at all.
I'm writing text, trying to counter the disgusting sadness and degenerate self hate that people like you naively believe in and are embodying, true, not much more in this context (but not the only thing I'm doing), did you ever think about what a word or even a symbol, sign or letter is ?
Guys. Arthur is fake. He can't answer any question specifically. He can't tell you the epistemic bases for his arguments, or even explicate what those arguments are. He can only make irrelevant ad-hominem arguments about his emotional state. His text screams "teen-ager who's in his Nietzsche phase."
Probably time to drop the thread or ignore Arthur.
-
ian807
- Tar Sands
-
- Posts: 899
- Joined: Mon 03 Nov 2008, 04:00:00
-
by Carlhole » Sun 02 Jan 2011, 19:20:21
$this->bbcode_second_pass_quote('ian807', 'h')ttp://bluebrain.epfl.ch/
The Blue Brain Project... My favorite computer science project. Just read the FAQ page!
Aside from reverse-engineering the brain and all that, the BBP wants to be set up as the dynamic online repository for everything that is known about the function of neurons and networks in living human brain tissue. Any qualified neuroscientist anywhere in the world could contribute further research to this Standard Model or could use the Standard Model to conduct experiments, design drugs, etc.
Meanwhile, the Indian mathematicians over at IBM BlueMatter insist that you do no have to first mimic nature to derive AI in a supercomputer. However, even this group is involved in ultra-hi resolution brain slicing and scanning into digital. We'll definitely be hearing more interesting news to come from all this.
-
Carlhole
-
by Arthur75 » Sun 02 Jan 2011, 19:54:45
Even birds are now disgusted by the giant airheads trepanated impotent pigs farm that is the only thing left of America :
http://www.youtube.com/watch?v=Y1ZrUYZgaHQ"America is the only country that went from barbarism to decadence without civilization in between. "
A bit sad ...
-
Arthur75
- Tar Sands
-
- Posts: 529
- Joined: Sun 29 Mar 2009, 05:10:51
- Location: Paris, France
-
by Carlhole » Sun 02 Jan 2011, 21:13:08
IBM’s “Watson” Computing System to Challenge All Time Greatest Jeopardy! Champions$this->bbcode_second_pass_quote('', '&')quot;After four years, our scientific team believes that Watson is ready for this challenge based on its ability to rapidly comprehend what the
Jeopardy! clue is asking, analyze the information it has access to, come up with precise answers, and develop an accurate confidence in its response,” said Dr. David Ferrucci, the scientist leading the IBM Research team that has created Watson.
“Beyond our excitement for the match itself, our team is very motivated by the possibilities that Watson's breakthrough computing capabilities hold for building a smarter planet and helping people in their business tasks and personal lives."
"We're thrilled that
Jeopardy! is considered a benchmark of ultimate knowledge,” said Harry Friedman, Executive Producer of
Jeopardy!. “Performing well on
Jeopardy! requires a combination of skills, and it will be fascinating to see whether a computer can compete against arguably the two best
Jeopardy! players ever."
...an example of an emulation.
-
Carlhole
-
by EnergyUnlimited » Mon 03 Jan 2011, 05:48:41
$this->bbcode_second_pass_quote('ian807', '')$this->bbcode_second_pass_quote('EnergyUnlimited', '')$this->bbcode_second_pass_quote('Article posted by Carlhole', 'H')owever, if we take the possibility of successful human brain emulations this century
seriously, and find some plausibility to the above discussion, more rigorous and extensive
analysis would be well worth the effort.
That is perhaps the most important bit.
We are dealing with academic speculation and not with credible ongoing project.
It is just theoretical
what if... discussion.
Um... No.
http://bluebrain.epfl.ch/ Nice high profile project designed to keep cash flow to academia going.
Very much like plethora of fusion projects, I suppose.
But what have they delivered up to date?
They are claiming some success with modeling of rat neurocortical column but they fail to indicate any practical use of invention.
We even don't know how accurate these emulations are... may be as accurate as
rat droppings used to emulate coffee... may be slightly more (or less) accurate...
Will they deliver anything useful in the future?
Probably, but it will be a far cry from what they claim.
That is under condition that cash flow will continue and this is far from certain.
Meantime Moore Law is falling apart... hardware development problems are looming... cash (including cash for academia) is draining out... and industrial societies are on the brink of retreat...
-
EnergyUnlimited
- Light Sweet Crude
-
- Posts: 7537
- Joined: Mon 15 May 2006, 03:00:00
-
by Carlhole » Mon 03 Jan 2011, 09:11:30
On Genes, Memes, Bemes, and Conscious Things$this->bbcode_second_pass_quote('', 'W')hen will be be able to beme ourselves up into transbeman states? Right now, mind files exist in a virtual reality on the Internet. People are creating more of themselves in websites such as secondlife.com. Richard Morgan has laid out a number of scenarios for
ex vivo consciousness. We cannot predict an exact date, but remember that only sixty years ago, Vannevar Bush predicted a memory extender machine that is available today in desktop, PDA, and ipod versions.
A bemex hypothesis to complement the memex hypothesis – a bemex is a device in which an individual stores enough of their bemes and which is equipped with mindware so that it may function as our alter ego. It is an analog of one’s consciousness that can be replicated with speed and flexibility. Kurzweil predicts the arrival of this bemex capability for the year 2030. We are certainly within the horizon of this and we should be debating the related ethical issues.
It isn't hard to imagine Hollywood eventually producing a completely virtual "movie star", complete with unique character that never ages, or who is available at any age for that matter. Such a "beman" would eventually have to be able to do an interview with Letterman though.
-
Carlhole
-
by Carlhole » Mon 03 Jan 2011, 09:27:07
On Genes, Memes, Bemes, and Conscious Things$this->bbcode_second_pass_quote('', 'W')hen will be be able to beme ourselves up into transbeman states? Right now, mind files exist in a virtual reality on the Internet. People are creating more of themselves in websites such as secondlife.com. Richard Morgan has laid out a number of scenarios for
ex vivo consciousness.
A bemex hypothesis to complement the memex hypothesis – a bemex is a device in which an individual stores enough of their bemes and which is equipped with mindware so that it may function as our alter ego. It is an analog of one’s consciousness that can be replicated with speed and flexibility. Kurzweil predicts the arrival of this bemex capability for the year 2030. We are certainly within the horizon of this and we should be debating the related ethical issues.
-
Carlhole
-
by ian807 » Mon 03 Jan 2011, 09:56:44
$this->bbcode_second_pass_quote('EnergyUnlimited', '')$this->bbcode_second_pass_quote('ian807', '')$this->bbcode_second_pass_quote('EnergyUnlimited', '')$this->bbcode_second_pass_quote('Article posted by Carlhole', 'H')owever, if we take the possibility of successful human brain emulations this century
seriously, and find some plausibility to the above discussion, more rigorous and extensive
analysis would be well worth the effort.
That is perhaps the most important bit.
We are dealing with academic speculation and not with credible ongoing project.
It is just theoretical
what if... discussion.
Um... No.
http://bluebrain.epfl.ch/ Nice high profile project designed to keep cash flow to academia going.
Very much like plethora of fusion projects, I suppose.
But what have they delivered up to date?
They are claiming some success with modeling of rat neurocortical column but they fail to indicate any practical use of invention.
We even don't know how accurate these emulations are... may be as accurate as
rat droppings used to emulate coffee... may be slightly more (or less) accurate...
Will they deliver anything useful in the future?
Probably, but it will be a far cry from what they claim.
That is under condition that cash flow will continue and this is far from certain.
Meantime Moore Law is falling apart... hardware development problems are looming... cash (including cash for academia) is draining out... and industrial societies are on the brink of retreat...
Of course, exactly the same argument could have been made for that other nice high profile project designed to keep cash flow to academia going (it was called "Arpanet") in 1969. It had been on the drawing board for about seven years or so by that time. Would it ever deliver anything useful in the future? Probably, but it turned out to be a far cry from what they claimed (It delivered wildly more than expected). These days, we call it "The Internet."
-
ian807
- Tar Sands
-
- Posts: 899
- Joined: Mon 03 Nov 2008, 04:00:00
-
by Dezakin » Mon 03 Jan 2011, 15:12:10
$this->bbcode_second_pass_quote('EnergyUnlimited', '')$this->bbcode_second_pass_quote('ian807', '')$this->bbcode_second_pass_quote('EnergyUnlimited', '')$this->bbcode_second_pass_quote('Article posted by Carlhole', 'H')owever, if we take the possibility of successful human brain emulations this century
seriously, and find some plausibility to the above discussion, more rigorous and extensive
analysis would be well worth the effort.
That is perhaps the most important bit.
We are dealing with academic speculation and not with credible ongoing project.
It is just theoretical
what if... discussion.
Um... No.
http://bluebrain.epfl.ch/ Nice high profile project designed to keep cash flow to academia going.
Very much like plethora of fusion projects, I suppose.
But what have they delivered up to date?
They are claiming some success with modeling of rat neurocortical column but they fail to indicate any practical use of invention.
We even don't know how accurate these emulations are... may be as accurate as
rat droppings used to emulate coffee... may be slightly more (or less) accurate...
Will they deliver anything useful in the future?
Probably, but it will be a far cry from what they claim.
That is under condition that cash flow will continue and this is far from certain.
Well, its a software project so it can continue on a budget much smaller than the fusion white elephants. It has huge mindshare, so you will have many pursuing it out of interest even if not financial motive. And the immediate financial motive is in drug development, which is... lucrative.
I'm actually siding with you that the future will have a number of unforeseen engineering problems that will render much of the simulation useless for any sort of AI. But that's not the same as agreeing these problems will be forever unresolvable. You have to assume that there are processes in the brain that are unsimulatable that are essential for intelligence, and frankly I find that very implausible.
$this->bbcode_second_pass_quote('', 'M')eantime Moore Law is falling apart... hardware development problems are looming... cash (including cash for academia) is draining out... and industrial societies are on the brink of retreat...
You're citing impending problems that don't exist yet. Moore's law on the observation of transistor densities at cost increases has continued to hold up, and will for another decade. There are challenges ahead certainly, but they're interesting ones with known solutions besides simply shrinking feature sizes. Self timed circuits, adiabatic logic, and 3d circuitry are all developed options. Reconfigurable computing offers enormous advantages as well.
The notion that industrial societies are on the brink of retreat is popular here, but the notion that the decline of fossil fuels will happen fast enough to eliminate civilization's capacity to innovate is suspicious.
-
Dezakin
- Heavy Crude
-
- Posts: 1569
- Joined: Wed 09 Feb 2005, 04:00:00
-
by EnergyUnlimited » Mon 03 Jan 2011, 16:32:20
$this->bbcode_second_pass_quote('Dezakin', '
')Well, its a software project so it can continue on a budget much smaller than the fusion white elephants. It has huge mindshare, so you will have many pursuing it out of interest even if not financial motive. And the immediate financial motive is in drug development, which is... lucrative.
Do you believe that existing or soon to come hardware is sufficient to run a very advanced software which is in all probabilities needed to run some useful emulations?
$this->bbcode_second_pass_quote('', 'I')'m actually siding with you that the future will have a number of unforeseen engineering problems that will render much of the simulation useless for any sort of AI. But that's not the same as agreeing these problems will be forever unresolvable.
If a feature (brain) exist, it is an evidence that in principle it can be made.
However I am deeply suspicious that such task is within reach during coming few decades.
$this->bbcode_second_pass_quote('', 'Y')ou have to assume that there are processes in the brain that are unsimulatable that are essential for intelligence, and frankly I find that very implausible.
They are making brain scans and presumably all what they can track now are patterns of electrical connections.
What about memory storage (which presumably works base on residual peptide sequences)?
What about mechanisms of retrieving data from this memory?
What about error handling procedures (I believe these are most important when it comes to creativity)?
There is much more in a neuron than electrical network and its I/O system.
Current tools are rather inadequate to investigate this.
Is it possible to resolve?
In principle obviously yes but we may well prove not to be smart enough.
On the top of it brain is not digital (we had a discussion about it sometime ago).
$this->bbcode_second_pass_quote('', '
')You're citing impending problems that don't exist yet. Moore's law on the observation of transistor densities at cost increases has continued to hold up, and will for another decade.
Red an article not long ago suggesting that coming round of chips is going to be expensive enough to kill all the profit due to manufacturing tooling expenses.
Will search for it again to give a reference I you want.
$this->bbcode_second_pass_quote('', 'T')here are challenges ahead certainly, but they're interesting ones with known solutions besides simply shrinking feature sizes. Self timed circuits, adiabatic logic, and 3d circuitry are all developed options. Reconfigurable computing offers enormous advantages as well.
I don't know what is adiabatic logic (have to search for it), however with 3D designs there is an obvious concern related to heat dissipation.
We are fast approaching hard limits of silicon technology.
In any case at the moment it seems that it is a software what is lagging behind.
While processors are getting faster, Windows is getting slower.
Thrown it away, work on Linux now.
$this->bbcode_second_pass_quote('', 'T')he notion that industrial societies are on the brink of retreat is popular here, but the notion that the decline of fossil fuels will happen fast enough to eliminate civilization's capacity to innovate is suspicious.
Don't you get any hint from pending monetary collapse?
We are approaching game over situation where academias are disbanded and computer scientists out of bare necessity to feed family keep busy themselves with repairs of TV-s or hair dryers.
Very much what have happened in ailing Soviet Union.
We discussed a lot about nuclear (fission) power and we seem to agree that it is a way forward to attempt to preserve whatever can be saved from our civilization.
Do you observe any crash programs to scale it up?
I don't.
It is just talk and not enough is done to even replace NPP which are about to go out of service.
And time is running out...
We are running out of cash, meantime FF are getting more expensive and sparser...
Meantime politicians are spending money they don't have on projects which society don't need (mainly banker welfare schemes).
-
EnergyUnlimited
- Light Sweet Crude
-
- Posts: 7537
- Joined: Mon 15 May 2006, 03:00:00
-
by Dezakin » Tue 04 Jan 2011, 01:22:01
$this->bbcode_second_pass_quote('EnergyUnlimited', '')$this->bbcode_second_pass_quote('Dezakin', '
')Well, its a software project so it can continue on a budget much smaller than the fusion white elephants. It has huge mindshare, so you will have many pursuing it out of interest even if not financial motive. And the immediate financial motive is in drug development, which is... lucrative.
Do you believe that existing or soon to come hardware is sufficient to run a very advanced software which is in all probabilities needed to run some useful emulations?
Yes; At least useful enough to do work with drug development if not any AI work. Existing hardware is certainly sufficient, especially using reconfigurable computing (FPGAs.)
$this->bbcode_second_pass_quote('', '')$this->bbcode_second_pass_quote('', 'I')'m actually siding with you that the future will have a number of unforeseen engineering problems that will render much of the simulation useless for any sort of AI. But that's not the same as agreeing these problems will be forever unresolvable.
If a feature (brain) exist, it is an evidence that in principle it can be made.
However I am deeply suspicious that such task is within reach during coming few decades.
I believe we'll be able to simulate something on the scale of the human brain in several decades, but that it won't actually operate the way a human brain does because of our current lack of understanding of the architecture of the human brain. Whale brains are many times the size of the human brain but less 'intelligent' because the architecture is different. How long it takes to go from human brain scale to intelligent architecture is something I'm not sure of but I'd guess it will take not less than ten years after getting the scale of the human brain in simulation and not more than forty years. Then after that you have to raise the brain in software, and our experience with human children indicates that's another ten to twenty years.
We could bypass that if we could scan in existing neural structures of adult human brains, but I couldn't guess how long that will take to mature. It will probably be simpler to raise an infant brain in software sooner than downloading an adult brain raised in the meat world.
So in my opinion the effects of brain simulation AI are at least four decades away, but not more than eighty years away.
$this->bbcode_second_pass_quote('', '')$this->bbcode_second_pass_quote('', 'Y')ou have to assume that there are processes in the brain that are unsimulatable that are essential for intelligence, and frankly I find that very implausible.
They are making brain scans and presumably all what they can track now are patterns of electrical connections.
What about memory storage (which presumably works base on residual peptide sequences)?
What about mechanisms of retrieving data from this memory?
What about error handling procedures (I believe these are most important when it comes to creativity)?
There is much more in a neuron than electrical network and its I/O system.
Current tools are rather inadequate to investigate this.
Is it possible to resolve?
In principle obviously yes but we may well prove not to be smart enough.
On the top of it brain is not digital (we had a discussion about it sometime ago).
The blue brain project requires the large hardware resources it does because it incorporates the NEURON package that does biochemical simulation of neurons, not just the network of connections that simplistic artificial neural nets utilize.
http://en.wikipedia.org/wiki/Computational_neuroscience
Computational neuroscience isn't standing still. It would be useless for doing neuroscience for medicine if it didn't simulate biological neurons. I'm sure we're going to run into problems with our model and find resolutions to more accurately model biological neural nets in digital systems.
When we develop brain simulation AI I imagine we'll be able to toss out all the extra computing time that goes into fully simulating biological neural nets and only focus on whats essential. We don't know what that is yet, so that is decades down the line.
$this->bbcode_second_pass_quote('', '')$this->bbcode_second_pass_quote('', 'T')here are challenges ahead certainly, but they're interesting ones with known solutions besides simply shrinking feature sizes. Self timed circuits, adiabatic logic, and 3d circuitry are all developed options. Reconfigurable computing offers enormous advantages as well.
I don't know what is adiabatic logic (have to search for it), however with 3D designs there is an obvious concern related to heat dissipation.
We are fast approaching hard limits of silicon technology.
We're approaching hard limits of bulk CMOS in reduction of feature size, but not yield or transistor density. Self clocked circuits can reduce power dissipation considerably, but require more engineering, enabling larger circuits with higher effective power densities. Reversible computing techniques can vastly reduce the power dissipation of circuits. Reconfigurable computing (FPGAs) offer orders of magnitude faster execution of some algorithms. There are advances I understand less, such as memristors, spintronics, and other applications of nanoelectronics that simply push the power dissipation problem down and the feature size down that I know are being developed.
The engineering for these problems are hard, so I expect another decade of bulk CMOS advances down to maybe 10nm before running into a wall and entering the new hard problem domain of nanoelectronics and harder engineering problems. Because these problems are that much harder I expect we'll see a slowdown in computational power at cost, but it won't halt. It will honestly be interesting to see techniques pursued that are more elegant than the brute force technique of throw more smaller transistors at the problem in a cruder fashion.
$this->bbcode_second_pass_quote('', '')$this->bbcode_second_pass_quote('', 'T')he notion that industrial societies are on the brink of retreat is popular here, but the notion that the decline of fossil fuels will happen fast enough to eliminate civilization's capacity to innovate is suspicious.
Don't you get any hint from pending monetary collapse?
Look, the thesis is that we're teetering on the brink of a collapse so total that we can't climb out of it I find implausible, especially with so much industrial capacity that is geared towards frivolous utilization of resources. We're just going to have to agree to disagree on this, because I find it unlikely that:
1) Massive collapse is imminent (or well defined. I don't know what massive collapse actually means here to most people)
2) Economic collapse due to energy shortage is unrecoverable given our vast capacity to restructure.
3) People are so disorganized and factional that global civilization will simply die rather than restructure.
-
Dezakin
- Heavy Crude
-
- Posts: 1569
- Joined: Wed 09 Feb 2005, 04:00:00
-
by ian807 » Tue 04 Jan 2011, 16:03:39
$this->bbcode_second_pass_quote('Dezakin', '
')Look, the thesis is that we're teetering on the brink of a collapse so total that we can't climb out of it I find implausible, especially with so much industrial capacity that is geared towards frivolous utilization of resources. We're just going to have to agree to disagree on this, because I find it unlikely that:
1) Massive collapse is imminent (or well defined. I don't know what massive collapse actually means here to most people)
2) Economic collapse due to energy shortage is unrecoverable given our vast capacity to restructure.
3) People are so disorganized and factional that global civilization will simply die rather than restructure.
1)
Massive collapse is imminent. Perhaps. Slow collapse is already starting if you ask the folks in Detroit. The near-term threat is economic (i.e. worldwide debt loads). This could be massive. Liquid fuels depletion is long-term, but no less insidious in the long run. Our current civilization is remarkably dependent on this energy dense, flexible material for which there is no adequate substitute in quantities that matter. Changeover to anything else won't happen quickly, cheaply or easily.
2)
Economic collapse due to energy shortage is unrecoverable given our vast capacity to restructure.Yes, we'll recover, but in what timeframe? How does a century or two sound? An energy shortage in liquid fuels that causes significant supply chain disruption will in turn disrupt other networks such as power distribution, food production and communications infrastructure. Our systems are interdependent. It takes communications, transportation and power to maintain and build the communications, transportation and power systems. Remove one piece and you slow down or stop the process of maintenance of these systems. Feedback ensues until a bottom is reached. This bottom could last quite a while.
3)
People are so disorganized and factional that global civilization will simply die rather than restructure.
People will reorganize activities on a lower-power, local scale, possibly for a long time.
-
ian807
- Tar Sands
-
- Posts: 899
- Joined: Mon 03 Nov 2008, 04:00:00
-
by EnergyUnlimited » Tue 04 Jan 2011, 18:27:38
ian807,
If you believe in coming crash how that can be reconciled with speedy AI development, as per your other predictions?
What the use such AI would have if we are likely to contend with more down to Earth existential issues perhaps in centuries to come?
-
EnergyUnlimited
- Light Sweet Crude
-
- Posts: 7537
- Joined: Mon 15 May 2006, 03:00:00
-
by ian807 » Tue 04 Jan 2011, 18:56:21
$this->bbcode_second_pass_quote('EnergyUnlimited', 'i')an807,
If you believe in coming crash how that can be reconciled with speedy AI development, as per your other predictions?
What the use such AI would have if we are likely to contend with more down to Earth existential issues perhaps in centuries to come?
If you believe in coming crash how that can be reconciled with speedy AI development, as per your other predictions?
It's a race. Technological progress is continuing, but not at the same rate everywhere. Moreover, when the crashes come (and they will be multiple), they won't occur at the same times and in the same places. Even as the USA declines, China and India are rising. Another few decades, and these powers may be declining while Africa and South America become dominant. There will be AI research going on
somewhere for quite some time. The payoff is simply too high for it to be ignored. AI is the single most profitable technology humans can be developing at this point, even if it doesn't pay off for decades. By solving the problem of AI, we solve all the other solvable problems as a side effect.
What the use such AI would have if we are likely to contend with more down to Earth existential issues perhaps in centuries to come?It gives us the answers to any and every problem for which there
are available answers, including, among other things, power generation and the practical means to get to sustainable power. It may also moot the power questions in some other way by providing completely unexpected things like anti-gravity, FTL travel, cheap hyper-efficient building and insulating materials, metabolic and genetic modifications that allow us to live with less food, water, heat, etc. We can't know what a scalable human-like AI would come up with.
-
ian807
- Tar Sands
-
- Posts: 899
- Joined: Mon 03 Nov 2008, 04:00:00
-
by EnergyUnlimited » Tue 04 Jan 2011, 19:18:21
$this->bbcode_second_pass_quote('ian807', '
')It gives us the answers to any and every problem for which there are available answers, including, among other things, power generation and the practical means to get to sustainable power.
We already know that it is nuclear and renewables mix together with conservation and of course population reduction and we don't need AI to tell us that.
So I don't understand why would we need AI to learn the obvious?
$this->bbcode_second_pass_quote('', 'I')t may also moot the power questions in some other way by providing completely unexpected things like anti-gravity, FTL travel,
AI still have to work within realm of laws of physics so it is naive to expect above.
$this->bbcode_second_pass_quote('', 'c')heap hyper-efficient building and insulating materials,
Cheap means common, available from readily available materials.
I am afraid, there is not much room for improvement here.
Basic chemistry delivering such building materials is really well investigated area.
$this->bbcode_second_pass_quote('', 'm')etabolic and genetic modifications that allow us to live with less food, water, heat, etc.
These are more likely to be achieved, as long as AI wish to cooperate on the subject.
However biological systems are pretty well fine tuned so it maybe difficult to get substantial improvements.
The same may be true with intelligence.
It may be difficult to the extreme to create AI more intelligent than we are.
We are already a product of long fine tuning and there may not be much scope for improvement.
$this->bbcode_second_pass_quote('', 'W')e can't know what a scalable human-like AI would come up with.
Of course we can't.
However it must stay within realm of laws of physics for sure.
What if we don't like some solutions offered by AI (say that we need to exterminate 90% of peoples if we want to hope for sustainable, high tech life?
In overall my impression is that you expect too much from AI.
To get there we would rather need not AI but some friendly God like characters.
-
EnergyUnlimited
- Light Sweet Crude
-
- Posts: 7537
- Joined: Mon 15 May 2006, 03:00:00
-
by ian807 » Wed 05 Jan 2011, 00:01:14
$this->bbcode_second_pass_quote('EnergyUnlimited', '
')
So I don't understand why would we need AI to learn the obvious?
LOL. You may have a point there. There is no shortage of solutions, however there is a shortage of solutions that people will actually implement. Since we're on the subject of obvious, we should have started prepping for peak oil in the 80s. Didn't happen.
"...within the realm of physics."The end of physics has been declared before. Then that pesky Einstein fellow came along. Quantum physics soon followed. Then came nuclear bombs and power. A 19th century man would have declared these "not within the realm of physics." I doubt we're done with physics yet, or even close.
Cheap means common, available from readily available materials.
I am afraid, there is not much room for improvement here. Basic chemistry delivering such building materials is really well investigated area.Yes, it is, for humans, which require food and sleep and have limited attention, working memory and intelligence. A scalable AI would have none of these limitations, and so would have a much larger search space of memory and time for solutions. Assuming human limitations for an AI, or assuming that ours do not matter is probably a mistake.
$this->bbcode_second_pass_quote('', 'm')etabolic and genetic modifications that allow us to live with less food, water, heat, etc.
These are more likely to be achieved, as long as AI wish to cooperate on the subject.An AI will do what we tell it to do. It has no motivations other than those we provide it. Even self-preservation is not a given. An AI is fundamentally alien since it did not evolve gradually from a self-replicator like DNA.
However biological systems are pretty well fine tuned so it maybe difficult to get substantial improvements.
The same may be true with intelligence. It may be difficult to the extreme to create AI more intelligent than we are.Fooey. If that were true, there would be no differences in physical abilities or intelligence between animals and people, or between different people. That's clearly not true. If intelligence can be made to vary in organic creatures, there's no reason it can't be made to vary in silicon or some other matrix.
An AI may indeed seem godlike to us, but I doubt that it will be either friendly or unfriendly. It would simply be a truth telling machine. I guarantee we won't like some of the solutions found by AIs any more than we like those proposed by humans. An AI might be even more devastating in that it would have the capacity to see us, and the world, far more objectively than we can. Humans react poorly to unadorned truth.
-
ian807
- Tar Sands
-
- Posts: 899
- Joined: Mon 03 Nov 2008, 04:00:00
-
by EnergyUnlimited » Wed 05 Jan 2011, 03:18:10
$this->bbcode_second_pass_quote('ian807', '
')The end of physics has been declared before. Then that pesky Einstein fellow came along. Quantum physics soon followed. Then came nuclear bombs and power. A 19th century man would have declared these "not within the realm of physics." I doubt we're done with physics yet, or even close.
We are in position to run very advanced astronomical observations and investigate this way most extreme objects present in Universe.
Yet no hint of FTL is found (so causality is preserved) and antigravity, if exists at all, takes a form of dark energy which can deliver measurable effects on a scale of galaxy cluster and above.
We are not going to play with these anytime soon.
I am afraid that our current understanding of laws of physics is pretty good down to Planck scale where our models brake down.
However Planck scale is and will remain inaccessible for us for good due to practical limitations.
We are not going to build Galaxy sized particle accelerator to deliver adequate energies and any smaller one would fall fool of synchrotron radiation.
$this->bbcode_second_pass_quote('', 'Y')es, it is, for humans, which require food and sleep and have limited attention, working memory and intelligence. A scalable AI would have none of these limitations...
But this is irrelevant from human perspective.
You was talking about production of bulk quantities of superb building/insulating materials.
AI would not need these (albeit it would need other resources like rare elements used in semiconductor tech) but humans do.
$this->bbcode_second_pass_quote('', 'A')n AI will do what we tell it to do. It has no motivations other than those we provide it. Even self-preservation is not a given.
I would not take these assumptions for granted.
$this->bbcode_second_pass_quote('', 'I')f that were true, there would be no differences in physical abilities or intelligence between animals and people, or between different people. That's clearly not true.
These assumptions are simply not true.
Animals have various strategies of survival and intelligence is only one of these.
Humans elected this path of evolution and they are pretty well fine tuned already.
Survival strategy of other animals rely on different factors and some of them might eventually outcompete us.
In respect of differences between humans, one can say that once you remove obviously pathological cases, intelligence difference between individuals is not as great as one can think.
Intelligence not necessarily translate into success as well.
Few years ago the most intelligent human being was an Italian housewife with IQ well above 200.
$this->bbcode_second_pass_quote('', 'I')f intelligence can be made to vary in organic creatures, there's no reason it can't be made to vary in silicon or some other matrix.
Of course it can.
My query is about possible upper limits achievable in practice.
$this->bbcode_second_pass_quote('', 'A')n AI might be even more devastating in that it would have the capacity to see us, and the world, far more objectively than we can. Humans react poorly to unadorned truth.
That is why they might not benefit of AI as much as you hope.
-
EnergyUnlimited
- Light Sweet Crude
-
- Posts: 7537
- Joined: Mon 15 May 2006, 03:00:00
-
by Arthur75 » Wed 05 Jan 2011, 06:33:05
$this->bbcode_second_pass_quote('ian807', 'H')umans react poorly to unadorned truth.
True, and your reaction when confronted to the truth that AI is nothing but a modern form of alchemy with associated clowns babble, promises, and other lies, is a good example of that.
-
Arthur75
- Tar Sands
-
- Posts: 529
- Joined: Sun 29 Mar 2009, 05:10:51
- Location: Paris, France
-
by NickyBoy » Wed 05 Jan 2011, 13:25:26
$this->bbcode_second_pass_quote('Arthur75', '')$this->bbcode_second_pass_quote('ian807', 'H')umans react poorly to unadorned truth.
True, and your reaction when confronted to the truth that AI is nothing but a modern form of alchemy with associated clowns babble, promises, and other lies, is a good example of that.
I'm sure many once said that about flight.
-
NickyBoy
- Peat
-
- Posts: 68
- Joined: Wed 04 Feb 2009, 04:00:00
-
Return to Open Topic Discussion
Who is online
Users browsing this forum: No registered users and 1 guest
