Asymmetrical Information: Say it ain't so, Joe

Call me a starry-eyed optimist but I have great faith that whatever health risks are posed by interplanetary travel will be overcome as they present themselves. The article dismisses ship-based magenetic fields out of hand but clearly the Earth's magnetic field does the trick. Why would it be impossible to reproduce this?

I'm not a scientist and shouldn't be taken as an expert on anything but it seems to me humanity has overcome much more difficult problems in the past and should be able to do so again.

Posted by: Christopher Ross on September 18, 2006 5:32 PM

I was kind of surprised by the suggestion that the moon would be a serious problem. It doesn't take that many feet of dirt to get something equivalent to the earth's atmosphere, and not that many more to get something equivalent to the earth's magnetic field. But there seems little doubt, to me anyway, that a trip to Mars with our current resources is less than reasonable.

I think we should be concentrating on the moon and only on the moon until we've built a colony of signicicant size and have walked far enough down the learning curve that the idea of building spaceships with five feet of water shielding them doesn't sound implausible.

Posted by: Mark Amerman on September 18, 2006 5:38 PM

Whenever somebody talks about stopping "all" cosmic rays, I twitch. The atmosphere and magnetic field of Earth doesn't stop all cosmic rays.

And microgravity disease is a boogeyman. People have been in microgravity for over a year, with cumulative time exceeding two years.

Posted by: Warmongering Lunatic on September 18, 2006 6:16 PM

Hmm. Some of the radiation damage to cells could be mitigated by the approach described by Aubrey de Gray in his SENS proposal. That would provide an interesting juxtaposition; people with enhanced longevity who live in space...

Posted by: ellipsis on September 18, 2006 7:07 PM

Christopher:

The main reason that ship-produced magnetic fields would be impractical is that the Earth's magnetic field is *huge* -- it has a significant effect even when you're tens of thousands of miles above the ground.

To give you an idea of how difficult it would be to generate a magnetic field the size of the Earth's: A few years ago, the move "The Core" came out, a bad sci-fi movie about the Earth's magnetic field going away. It sparked a Usenet discussion about generating an artificial magnetic field around the Earth, and whether such a thing would be plausible.

You can make a magnet by running electricity through a wire wrapped around an iron core. Theoretically, if the Earth's natural magnetic field were to go away, we could wrap a gigantic copper wire around the planet and run an electric current through it.

The wire would need to be about 100 yards in diameter, and getting the electric current going would take 100 years of all of the output of all of the nuclear plants in the world. Once you had it going, it could be maintained with only a few nuclear power plants.

If the wire ever broke, there would be a spark, like you sometimes see when you unplug an electric device. That spark would be something on the order of a ten gigaton explosion -- about 500,000 times the Hiroshima bomb.

It'd be much easier to just plate your ship with a few inches of aluminum, lead, and iron.

Posted by: Aric on September 18, 2006 8:30 PM

Aric, Lead and Iron are the absolute worst materials to coat a spaceship with. You want to minimize the effects of radiation, and since Lead and Iron are such dense materials, it takes less thickness of those materials to stop primary radiation. However, using these dense materials also means that there is less thickness available to stop secondary radiation, the cascade of particles that results when a cosmic ray or particularly energetic solar particle hits your radiation shield.

Instead, radiation shields for interplanetary spacecraft should be made of the same mass of less-dense materials such as parrafin wax and water. That way, the secondary particles produced are much lighter and can be stopped by the increased thickness of the radiation shield.

Posted by: Ed Minchau on September 18, 2006 8:45 PM

What we need is speed. Instead of shoveling bundles of million dollar bills down the metaphorical rat-hole of near-earth-orbit we could instead begin a Manhattan style project to develop very high thrust non-chemical spacecraft engines. At our current R&D on such systems it'll be well into the next century before we have engines that will power spacecraft around the solar sytem in weeks and months instead of years and decades. Speed is of the essence. It doesn't merely reduce radiation exposure but it also greatly reduces the need for all manner of consumables such as oxygen, food, water, etc. which for even a modest crew size would weigh many tons for years long journeys.

A journey to Jupiter in a space craft that could accelerate continuously at one gravity could make the trip, including slowing down for Jupiter orbital injection, in mere weeks. Unmanned ships might be able to operate at multiple gees reducing resupply and equipment delivery times for far flung outposts to just a few days. This would be an immense technological challenge to put it mildly but I fear the solar system will be essentially beyond our reach and ability to learn of and exploit until we rise to it. Baby steps will get us nowhere slowly.

Posted by: M. Mitchell on September 19, 2006 1:10 AM

Will scanners live in vain? Maybe, maybe not.

Posted by: Joan on September 19, 2006 1:28 AM

The correct solution is the one that got dismissed in the article as "impractical" -- surround your ship with 5 feet of water, encased in the hull.

Whoever does it first will own space; whoever listens to these can't-do people will sit and watch somebody else own space.

Sheesh.

Posted by: Erich Schwarz on September 19, 2006 3:33 AM

Aric: The idea of making up the Earth's magnetic field artificially is fascinating. A couple of questions suggest themselves:

1) As you ran the electromagnet, would it gradually magnetize the planet's ferromagnetic core? This is at least plausible. Then, by the time you got it revved up (100 years or so), you could start winding it down.

2) Wouldn't this work for Mars?

Posted by: sammler on September 19, 2006 6:19 AM

Planning to send people into space before thechnology is ready is a waste of resouces. Forty years ago we had a crash program to go to the moon, we got there and and nothing came of it. The building of the shuttle was premature and it can not reach the synchronous orbit altitude, where it would be most useful. Almost all the valuable things from the space program are from unmanned missions or shuttle missions that could be unmanned if the shuttle did not exist. We need some cost benefit analysis before even thinking of starting a Mars program.

Posted by: joan on September 19, 2006 7:12 AM

Three Words: Von Neumann Probes.

Posted by: Paul Snively on September 19, 2006 11:04 AM

You want a space ship that is massive enough to have a five foot shield of water around it and that is far faster than any space craft we can put into the skies today? Easy. Build an Orion ship.

Posted by: tcobb on September 19, 2006 11:20 AM

Eric,
Surrounding the ship with 5ft of water is not the easy solution you envision. Assuming that the Mars craft would be roughly the same size as Apollo (1400sq ft of surface area not including engines) 5ft of water would weigh close to 250,000 tons. Add that to the ship and crew supplies and you have a ship that can't get off the ground with current technology. To assemble in space you'd need 5 shuttle trips just for the water. This ignores the constant steam and ice transitions that the water (and container) would need to endure in the heat and cold of space. As Mitchell said, speed is the only answer.

Posted by: CuriousTexan on September 19, 2006 11:36 AM

CuriousTexan, you're assuming the water needs to be brought up from Earth. Why not locate a chunk or two of floating ice out in the asteroid belt, and bring the water in from there?

(I really do read too much sf.)

Posted by: Joan on September 19, 2006 11:41 AM

I think that's more like 250 tons, not 250,000 tons.

Anyway, doesn't speed bring its own problems? Hitting a microgram particle at 0.01c is like taking a potshot from a .458 magnum.

Posted by: AT on September 19, 2006 11:50 AM

Speed and a water-shell aren't mutually exclusive solutions; build an Orion ship and you can have both.

Chewing gum and walking at the same time is an option.

Posted by: Erich Schwarz on September 19, 2006 12:49 PM

Aric,
Of course, I assumed any answers to these problems would be non-trivial. I mostly took issue with the article's enternal doom-and-gloom theme. While it may be possible that we are trapped on Earth for the next 50 or 100 or 500 years, there is no reason to believe technology won't find a way around it.

Does anyone know the effectiveness of dirt against cosmic rays?

Posted by: Christopher Ross on September 19, 2006 1:44 PM

"This ignores the constant steam and ice transitions that the water (and container) would need to endure in the heat and cold of space."

Good, since they're pretty much nonexistent. "Space" turns a spaceship into the world's largest Dewar flask. (The most common US tradename for Dewar flasks is Thermos.)

Posted by: Doug Sundseth on September 19, 2006 2:43 PM

AT--You're correct 250 tons
Joan--How do you protect the ship durrinng the mining operation?
Erich--Nice to know that someelse also knows of the Orion program. A nuclear bomb drive might work well in space, but I lack the imagination to see how you could use it for lift off. That's an awful lot of fall out.

Posted by: CuriousTexan on September 19, 2006 3:00 PM

It is possible to concoct scenarios whereby we could get humans to other planets in our solar system. An unanswered question is why we'd want to do that -- it would make a great deal more sense to build massive numbers of orbital habitats instead, using material from the asteroid belt. Mars will never be a decent place for humans to live, and as long as you're stuck living in a sealed environment it might as well be one closer to home.

As for colonizing other star systems -- there are ways to possibly do it (someone mentioned Von Neumann probes), but none of them involve sending actual living humans.

Posted by: Dan on September 19, 2006 3:05 PM

I offer no solutions, only optimism. Ways will be found to minimize the radiation risks.

Posted by: Adrian Reilly on September 19, 2006 3:18 PM

Nice to know that someelse also knows of the Orion program. A nuclear bomb drive might work well in space, but I lack the imagination to see how you could use it for lift off. That's an awful lot of fall out.

No more than any one of the hundreds of atmospheric tests conducted from the 40s through the 60s. But yes, Congress wouldn't approve it. That's too bad, because there's no other known technology that's good for launching massive payloads to LEO except chemical rockets. Everything would have to go up in several dozen Saturn V or Energia launches. $$$$.

Anyone up for fission fragment or nuclear salt water?

Posted by: AT on September 19, 2006 3:28 PM

Mars will never be a decent place for humans to live, and as long as you're stuck living in a sealed environment it might as well be one closer to home.

Except that Mars has gravity, atmosphere, rotation, water, and fuel. I don't think I could get used to living somewhere with an escape velocity of walking speed.

Posted by: AT on September 19, 2006 3:30 PM

By the time long-term space travel like this becomes a possibility rather than a pipe dream, we should have an operational space elevator which mitigates the greater problems of getting materials into orbit, right?

Posted by: Joan on September 19, 2006 3:45 PM

CuriousTexan,

My guess is that we need to push weapons labs to develop laser-ignited nuclear fusion of deuterium. If we have that, we should be able to use pure fusion detonations with no fission to activate them, which in turn would let us do a launch without the heavy-atom radioisotopes that people worry about from fallout.

I know Livermore et al. used to be working on that, but that may have been de-emphasized during the Somnolent Nineties.

More generally, there's a political contradiction between building effective spacecraft and keeping the U.S. from "militarizing space" -- any ship competent to get humans to Mars and back, alive, is ipso facto a powerful military asset as well, so we can't really build the ships without mortally offending the arms-control faction.

Posted by: Erich Schwarz on September 19, 2006 4:41 PM

I like the idea of orbital habitats. I still have my membership card from the L5 society of 25 years ago.

Posted by: Rex on September 19, 2006 6:17 PM

Simply being in orbit with spare delta-V and a rock large enough to survive re-entry is ipso-facto ownership of a powerful military asset.

I don't know how much delta-V a typical satellite carries; for a scary scenario picture someone gaining access to control of a satellite with enough delta-V in it's thrusters to deorbit. Depending on it's re-entry survivability, that could make a mess of whatever it hits.

Posted by: Ian Argent on September 19, 2006 8:31 PM

AT-Anyway, doesn't speed bring its own problems? Hitting a microgram particle at 0.01c is like taking a potshot from a .458 magnum.

Dead on. Space isn't a perfect vacuum. (You can 'hear' things in space). And at high speeds, space becomes abrasive.

Doug Sundseth suggested that a spaceship was like a thermos. But I don't think a vacuum would help protect you from large amounts of EM radiation, which wouldn't need air to propagate.

Can anyone here explain why nobody's talking about mass drivers? They seem to be a pretty logical solution. Given the amount of energy required to send a plane to cruising altitude, I'm surprised they aren't used already to save on fuel for take-off since launching a plane that way would give you altitude without the drag caused by a normal airplane wing, at least for a breif period.

If it comes down to it, human's DNA repair mechanisms are pretty bad. Experiments on bacteria have shown that these mechanisms could be improved with sufficient selective pressure. (Why they're they way they are currently is up for speculation. Defense against viruses, perhaps?)

Posted by: Ryan on September 19, 2006 9:24 PM

One thing that's alway amused me immensely is the argument by many space cadets (including Stphen Hawking, who -- to put it mildly -- should know better) that the human race has to colonize the Solar System so that somebody will survie if humanity all gets wiped out on Earth by either an asteroid strike or a nuclear-biological world war.

Regarding the latter: rest assured that no matter where in the Solar System humans go, we'll take our suicidal little toys with us -- especially the biological ones. (It is, after all, tremendously easier to send them to various place in the Solar System than it is to send US there.) And as for an asteroid strike: a growing number of SF writers are pointing out that while it will be tens of millions of years before the next Dinosaur Killer hits Earth from NATURAL causes, once we achieve the ability to start shoving asteroids around in order to divert them from Earth it will be a few thousand years at most before some political faction or nut cult deliberately crashes one INTO Earth.

Posted by: Bruce Moomaw on September 19, 2006 9:49 PM

sammler:

I don't really know the answers to your questions -- 1) magnetism isn't my strong suit, and 2) it's a half-remembered Usenet discussion from 5+ years ago.

Posted by: Aric on September 20, 2006 12:15 AM

Bruce Moomaw,

You seem bitter and I'm wondering why. Why the hostility?
Where does it come from? Do you know why you're angry?
Because I don't see anything here that merits it even
if your reasoning is valid.

Looking at that reasoning, I don't think it's correct.
Yes, it's true, if we have people living on the moon, or
in orbit around the earth, or on mars, that they could,
if the urge took them, do terrible things to the earth.

But funny thing if you have people living on earth, they
can do terrible things to the earth too.

People are just plain dangerous. Having them live only
on earth doesn't somehow make us safe.

The earth is a big place and it's a small place at the
same time. Humanity might be wiped out in any number of
ways. Most of the natural mechanisms are in some sense
inevitable since they are highly likely to occur in the
long run. On the other hand the odds of such a thing
happening in any given year are extremely low.

Far more immediate is the threat of human action. At this
point I see two practical mechanisms and I'll skip the
details. Both are of a nature that if a group deliberately
set out to kill all of humanity, they could probably do it.

The more likely course of human extinction is though
the unanticipated consequence of using certain weapons
during a war. Word wars tend to get out of hand.

Unless someone is deliberately trying to wipe out all
humanity then human settlements off the earth would probable
survive (and seed the resettlement of earth).

Of course we are very long way from being able to make
such settlements. We might if we try be able to construct
a colony on the moon this century. But such a thing is
of a no value as a safe haven unless this colony can live
independently. That might take centuries of development.
It might even take millenia.

Once one independently self-sufficient colony exists then
the whole solar system follows soon thereafter, because
once you're off the earth, one place is pretty much the
same as another.

There comes a point where even a malignant power would find
it improbably difficult to kill all life.

A pessimist would say we don't have centuries. Could be, but
raw as the deal may be I think it's the only game in town.

Posted by: Mark Amerman on September 20, 2006 5:12 AM

A few thoughts:

-- Mass Drivers: IIRC, although these are quite energy-efficient, the reason they're not useful for manned space launch is the extreme accelerations to which any payload is subjected during launch.

-- Kinetic-Energy Weapons: Ian, Jerry Pournelle has discussed this concept extensively on his website & elsewhere, under the name "Thor".

-- Laser Launch: As an alternative to mass drivers & space elevators, laser launch is also a possibility.

Posted by: Ming on September 20, 2006 8:00 AM

Why is it that every time I get ready to post something on this thread, someone else has already posted it? Project Orion, "Thor" and now laser launch (which has been tested on a small scale at White Sands Missle Range in the last three years), all my topics are belong to youse.

Well, I'll get pedantic, then. Nobody is going to live on the Moon, they will live in the Moon. Although the number of Moonquakes that occur on a regular basis is going to make that more interesting than previously thought...

Posted by: ellipsis on September 20, 2006 10:48 AM

Yes, ellipsis, but will they be baking helium-3 out of the regolith?

Posted by: AT on September 20, 2006 11:20 AM

You can get all that water up there at a reasonable price if you build a "Space elevator", too. It's not clear as yet whether this is just really, really hard, or so hard that by the time we can do it the problem has been solved some other way.

People are working on it, though...

Posted by: Mike Earl on September 20, 2006 11:27 AM

AT queried:

Yes, ellipsis, but will they be baking helium-3 out of the regolith?

That's what teleoperated equipment with weak AI is for. Think of all the overage adolescents who live with a controller in their hand, staring at a screen; what great lunar excavator/mining equipment operators they would make...

Posted by: ellipsis on September 20, 2006 1:27 PM

Except that Mars has gravity, atmosphere, rotation, water, and fuel.

Well, it has rotation and water, anyway.

Its gravity is only .37 of Earth's, which means that any humans living there long-term will suffer an impressive array of health problems. A habitat, in contrast, could produce Earth-normal articifical gravity by rotating.

Its atmosphere is freezing, poisonous, and only slightly better than a complete vaccuum (pressure is 0.008 that of Earth). Hard to see the advantage, there -- it would need terraforming techniques indistinguishable from magic to get it to a breathable, livable state.

Mars does have rotation, but I'm not sure why that is an advantage. Besides, getting orbital habitats to rotate is easy.

Mars also has water, but finding water in space is easy as well. Dragging a few icy asteriods or comets into Earth orbit would solve that problem.

As for fuel, that can be obtained from water; see above. There is also not much need for fuel if you're not trying to get into and out of a gravity well with chemical thrusters; for space-based propulsion, an Orion-style system using nuclear fusion would be better. For travel to and from planetary surfaces, space elevators are the way to go.

Posted by: Dan on September 20, 2006 5:09 PM

If, and it is always a big "if", our current technological development continues long enough, then colonizing the solar system and the rest of the galaxy will happen. The problems outlined in the Discover article will be overcome by either finding a protection against them, or by developing "humans" that are either immune to them, or capable of living with them. And one must remember that the definition of "human" will eventually include descendents that are not biological, but rather mechanical in nature.

Of course, we must survive our technology, survive the local galactic exterminators (asteroids, comets, and nearby supernovae). Once we become galactic-wide citizens, then we need to find a way to survive the death of the universe.:~)

Posted by: Yancey Ward on September 20, 2006 5:21 PM

I do not believe in mechanical intelligences. The possibility of increasing enhancement of biological intelligence is another thing, and likely to be seen in some crude fashion over the next 10 to 15 years, but mechanical? No, not a chance.

Posted by: ellipsis on September 20, 2006 7:53 PM

I do not believe in mechanical intelligences. The possibility of increasing enhancement of biological intelligence is another thing, and likely to be seen in some crude fashion over the next 10 to 15 years, but mechanical? No, not a chance.

When you say you don't believe in mechanical intelligences, what do you mean by "mechanical"? Do you mean "deliberately constructed", or are you referring to silicon brains vs. meat brains?

In any event, we already have computers that are smarter than almost the entire animal kingdom, as well as computers that can out-think humans in a host of different categories. And while computers are ill-suited for imitating the human brain's heavily-interconnected, hormone-addled, quasi-analog nature, it isn't clear that the way the human brain does things is the best way to achieve higher intelligence.

Posted by: Dan on September 20, 2006 9:29 PM

Dan asked:
When you say you don't believe in mechanical intelligences, what do you mean by "mechanical"?

I mean whatever the person I replied to meant...

When you say you don't believe in mechanical intelligences, what do you mean by "mechanical"? Do you mean "deliberately constructed", or are you referring to silicon brains vs. meat brains?

Oh, ok, I'll be more specific. I don't believe in deliberately constructed intelligences, based on what I've seen so far. Nor do I believe in silicon brains, becaue the complexity of the operation of what you refer to as "meat brains" seems to keep increasing as we learn more. It isn't unreasonable to suspect that there are quantum effects involved in human brains and thus in intelligence, which makes things really interesting if you ponder it.

In any event, we already have computers that are smarter than almost the entire animal kingdom, as well as computers that can out-think humans in a host of different categories.

Define "smart", please. The AI I'm aware of is marginally better at some pattern recognition than a housefly, but sure doesn't come in a package that can fly around the room...

And while computers are ill-suited for imitating the human brain's heavily-interconnected, hormone-addled, quasi-analog nature, it isn't clear that the way the human brain does things is the best way to achieve higher intelligence.

Human intelligence is the only model we have at present. AI proponents, like fusion power fans, are forever predicting a really big event some 20 years in the future. Minsky was an advisor to Kubrick's "2001" movie way back in 1968, and he stated at the time that "HAL-9000" was bound to happen, he just wasn't sure if it would take all 30+ years before 2001 to get there. In the late 1980's, the back-error-propagation algorithm burst onto the scene, and strong AI was just bound to be a reality in 20 years. So now it's nearly 20 years later, and back-prop is still one of the best methods around, but it's seriously limited in application.

Genetic algorithms & fuzzy logic were supposed to be worldbeaters as well, but once again they are limited in scope. Powerful tools, especially fuzzy logic in nonlinear control system applications, like autofocus cameras & such things, but limited.

The current MRI and PET studies of the human brain might reveal, in time, something that the AI community can use. But based on the show so far, I don't believe it.

Posted by: ellipsis on September 20, 2006 9:51 PM

It isn't unreasonable to suspect that there are quantum effects involved in human brains and thus in intelligence, which makes things really interesting if you ponder it.

It is, however, unreasonable to suspect that there's something uniquely biological about those possible quantum effects.

Define "smart", please.

Able to apply cognitive processing to solve a problem. Most animals don't do a lot of thinking.

The AI I'm aware of is marginally better at some pattern recognition than a housefly, but sure doesn't come in a package that can fly around the room...

I'm not sure what you mean by "better than a housefly". Houseflies don't recognize much of anything, which is why they fly into bug zappers or spend twenty minutes trying to fly through a glass window. Computers are much better at actually identifying what they are seeing, and have been for many years.

But yes, the housefly is smaller -- that's a separate issue unrelated to intelligence, though.

Human intelligence is the only model we have at present

That statement is only true if you make it a tautology -- i.e., if you first posit that humans are the only intelligent things that exist. If you accept the idea of animal intelligence, however, the fact that there are computers who can outthink animals makes it hard to avoid facing the fact that machine intelligence already exists. It just isn't as flexible as human intelligence.

But still, there are plenty of areas where it is superior. To name just one trivial example, I seriously doubt there's a human alive who could do a better job than Amazon's expert systems at guessing what books, movies, and music I'll like. Amazon frequently does a better job at guessing what I'll like than I myself do, for that matter, and I'm me. Sure, you could argue that a human with access to all that data could make even better predictions -- the problem, though, is that no human *could* access all that data.

back-error-propagation algorithm [...] Genetic algorithms & fuzzy logic [...] The current MRI and PET studies of the human brain might reveal, in time, something that the AI community can use

We're really talking about two different things, here. The history of AI research was, for a long time, dominated by attempts to mimic the human brain and the way humans think. That task did, indeed, turn out to be a heck of a lot more involved than expected. But that's not the only kind of AI. Minsky envisioned a computer with human-level intelligence; what we actually got was computers that are orders of magnitude smarter than humans within their limited area of expertise. The number of things that humans do better than computers decreases with each passing year -- and a good percentage of those things are things computers don't especially *need* to do. For example, computers are mediocre at emulating human senses, but humans are even worse at perceiving and comprehending the "environment" that software works in, unless that software is specifically coded to be easily understood by us (and even then it can be tricky).

Posted by: Dan on September 20, 2006 11:04 PM

I wrote:
It isn't unreasonable to suspect that there are quantum effects involved in human brains and thus in intelligence, which makes things really interesting if you ponder it.

Dan replied:
It is, however, unreasonable to suspect that there's something uniquely biological about those possible quantum effects.

I agree. So? Where did I ever make such a claim?

I asked:
Define "smart", please.

Dan replied:
Able to apply cognitive processing to solve a problem.

Define "cognitive" and "processing" and "problem". Is it "cognitive processing to solve a problem" if I wire up some NAND gates to perform complex arithmetic operations, are those NAND gates engaged in applying cognitive processing to solve a problem whenever I change the inputs? Or are they simply a collection of transistors in a certain configuration that generates elements from a predetermined output set in response to elements from an input set?

Most animals don't do a lot of thinking.

How do you know that? Is it because they don't post to this weblog, or some other reason? How do you know I'm not an animal or an AI?

I wrote:
The AI I'm aware of is marginally better at some pattern recognition than a housefly, but sure doesn't come in a package that can fly around the room...

Dan replied:
I'm not sure what you mean by "better than a housefly". Houseflies don't recognize much of anything, which is why they fly into bug zappers or spend twenty minutes trying to fly through a glass window.

Houseflies are good at identifying the kinds of threats that evolution has fitted them to recognize, and reacting to those threats by flying away, to pick one example. Some of the back-error-prop algorithms could react to a similar threat, given the right input sensors, in a shorter time than the housefly could. But the housefly didn't require tens of millions of training cycles, as soon as it can fly, it's ready to evade those threats. See the difference?

Computers are much better at actually identifying what they are seeing, and have been for many years.

Define "see", and tell what it is that computers "see". If I show a copy of La Gioconda to a computer, will it "see" the picture?

Dan:
But yes, the housefly is smaller -- that's a separate issue unrelated to intelligence, though.

Not unrelated to applied intelligence...

Me:
Human intelligence is the only model we have at present

That statement is only true if you make it a tautology -- i.e., if you first posit that humans are the only intelligent things that exist. If you accept the idea of animal intelligence, however, the fact that there are computers who can outthink animals makes it hard to avoid facing the fact that machine intelligence already exists. It just isn't as flexible as human intelligence.

Hey, if I define any term broadly enough, it can mean what I want it to mean. But by then, it may have become without meaning. And didn't you claim that animals don't think much, so how can there be (by your implied definition) "animal intelligence"? What does it mean to claim that computers can out-think animals?

Dan:
But still, there are plenty of areas where it is superior. To name just one trivial example, I seriously doubt there's a human alive who could do a better job than Amazon's expert systems at guessing what books, movies, and music I'll like. Amazon frequently does a better job at guessing what I'll like than I myself do, for that matter, and I'm me. Sure, you could argue that a human with access to all that data could make even better predictions -- the problem, though, is that no human *could* access all that data.

So a very large filing cabinet, with a means for cross referencing information via simple rules, is "thinking"? Does my car "think", too?

Me:
back-error-propagation algorithm [...] Genetic algorithms & fuzzy logic [...] The current MRI and PET studies of the human brain might reveal, in time, something that the AI community can use

Dan:
We're really talking about two different things, here. The history of AI research was, for a long time, dominated by attempts to mimic the human brain and the way humans think. That task did, indeed, turn out to be a heck of a lot more involved than expected. But that's not the only kind of AI. Minsky envisioned a computer with human-level intelligence; what we actually got was computers that are orders of magnitude smarter than humans within their limited area of expertise. The number of things that humans do better than computers decreases with each passing year -- and a good percentage of those things are things computers don't especially *need* to do.

What things do computers *need* do to? Do they need to swap out disk drives, or maintain their own environment? Do they need to create themselves? Do they need to feed themselves? Just curious, how does one measure what a computer needs, and how does the computer know what it needs?

Dan:
For example, computers are mediocre at emulating human senses, but humans are even worse at perceiving and comprehending the "environment" that software works in, unless that software is specifically coded to be easily understood by us (and even then it can be tricky).

These are true statements. But I fail to see how the PowerPC in the laptop I'm typing on is "thinking" or "intelligent" just because the collection of transistors within it is specifically designed to react in certain predetermined ways to binary inputs. By that standard, any old calculator is "thinking", too, because it's faster at interpreting certain binaries generated by keystrokes than I am.

But the calculator doesn't know what it is doing, it has no self awareness. Isn't self awareness one aspect of intelligence?

Posted by: ellipsis on September 21, 2006 10:49 AM