Slashdot Mirror
The Singularity Blinds Sci-Fi
foobsr writes "Popular Science has an article discussing the growing difficulties that Sci-Fi writers encounter when it comes to extrapolating current trends. Doctorow and
Stross , both former computer programmers, are rated to be prototypes of a new breed of guides to a future which due to
Vinge's Singularity might not happen for humanity once a proper super-intelligence - maybe as a Matrioshka Brain - has been created."
Also worth investigating is John C. Wright's The Golden Age Trilogy...I had bought the first book to read on my Vegas trip (honeymoon) next week. Already ripped through it. Set 10,000 years in the future, where the Singularity, if it hasn't already happened, is damned close.
The books (in order) are:
* The Golden Age
* The Phoenix Exultant
* The Golden Transcendence
That said, the first 50 pages of the first book are a little tough-going, given that Wright is painting a really alien picture and forcing you to catch up with his terminology, but in the end, it's worth it. Having just started the second book, I can tell you that one of the major themes is socialism vs. libertarianism, and as a subset of that personal responsibility to society.
blog |
because the vast majority of science fiction has always been "lets take present day concious and subconcious fears and talk about them in metaphors set into a future so that we can discuss them without censorship or fear"
On the other hand there is a minority of good, hard, scientific science fiction like Larry Niven.
In the year 3004 (assuming humans still exist) the vast majority of the human race will still be assholes, and if their personalities are downloaded into sugar cube sized computers they will be assholes with even less grip on reality that todays breed of assholes.
I think I am going to patent a method for inflicting virtual pain / beatings / torture / death on these future embedded personalities, because it will be the only way to keep the bastards in line.
A E Van Vogt wrote a great novel, The Anarchistic Colossus, which dealt with the issues of advancing technology vs human minds extremely well, thoroughly recommended, despite the fact that it is 20 or 30 years old there are many things in there that todays slashdot reasers will recognise as current actual concerns.
http://slashdot.org/~GuyFawkes/journal
Since the 70s, scientists and sci-fi authors have been promising that a revolution, including real AI, is "just around the corner". But the elusive breakthroughs recede further into the distance the more progress is made.
There are plenty of contemporary sci-fi authors working in the near-future, the next few decades or centuries, Alastair Reynolds, Richard Morgan and Neal Asher being among the most notable. Reynolds in particular is very good - his future humanity colonizes the stars using a mix of cryogenics and relativistic time, no warp drives here.
Also, he mistakes the point of pedandtry. No-one is bothered if the science is possible (yet) but any author worth his salt knows that the fictional technology must be CONSISTENT. A device can't act one way in one story and a completely different way in another, because if that happens, it's not sci-fi anymore but pure fantasy (and not even good fantasy). Sheer laziness and lack of talent on the part of the author.
Excellent points. The best science fiction writers (IMNSHO) are the ones that extrapolate the future based on human behavior and motivations, rather than where we think our technology will take us. Good science fiction is not about predicting tehnological advances. It should read like non-fiction that hasn't occured yet. My four favorite science fiction writers are Dick, Gibson, Stephenson, and Bester. Their novels have aged well, and seem to portray a pretty accurate picture of humanity's future because they all realize one thing: people do not change. Technological advance and trends aside, we are not that different from people thousands of years ago. Books like Do Androids Dream of Electric Sheep?, Neuromancer, Snow Crash, or The Demolished Man seem more and more likely, because the technological advance therorized on are secondary. We identify with the characters in books like these. These books address religion, corporate greed, politics, race relations, the military, etc. They seem plausible because the characters in these books act like we would. A good science fiction writer needs to make a few good extrapolations on where technology might be in the coming decades (nanotechnology, cloning, genetic modifications, interplanetary travel, worldwide computer networks, whatever), but the real value is addressing the human factor. A hundred (or a thousand) years from now, people will still be bitching about the government, religion, and corporations. We will still be greedy and giving, petty and generous, cruel and kind. Human beings do not change. When writing science fiction, it is important to retain that insight into human nature if you want accurately forecast where we are going.
However, the article is referring to a particular kind of science fiction (sometimes called "hard" SF) which is based upon realistically extrapolating current technology and trends into the future.
The problem is that reasonable extrapolation along a number of pathways leads to a future that is so alien that it is difficult to imagine, and even more difficult to think of anything to write about that would be entertaining to modern readers. The problem, is that humanity as we know it may not exist for much longer.
However, both Vinge and Stross have found literary ways around the singularity. Sort of the science fiction equivalent of "Left Behind." That is, even if the singularity occurs, it might not take everybody.
In any case, regardless, I recognize the possibility of non-humanlike AI, but then we enter into the realm of unquantifiable BS. How do we measure modelling, problem-solving and creativity abilities (other than by something that ends up looking shockingly like a Turing test?). What do those words mean outside of the human context? As I pointed out in another post, outside of very limited, constrained problem domains, we don't have any idea how to wire something up that can do even sub-human "problem-solving" or "modelling". The field of AI has provided lots of great algorithms that turn out to do a decent job at doing near-human-quality work in very limited domains, or much-less-than-human-quality work in slightly less limited, but still very constrained domains. The field of consciousness research, which aims to understand and presumably, eventually, model the human brain is still nascent.
I trust the instinct of Francis Crick who spent the last years of his life working on this problem that it will be a huge problem that dogs science for years to come. Just like how Einstein spent his last years looking for a TOE - guess what, here we are decades later, and we are _slightly_ closer, but basically up against a brick wall.
I recognize the ability (in theory) to self-improve or evolve rapidly in software would make a "Singularity" type of scenario at least conceivable (assuming there are no other barriers to this sort of rapidly improving digital intelligence) if you can get past the humongous hurdles in getting there. I just don't think it's likely to happen in the next 10 or 20 or 30 years. And beyond that, I prefer not to speculate, or at least not to pretend that my speculations are much more than pure science fiction themselves.
Yes, I should have included some kind of qualifier for speed. I do realize that properly designed human-emulation software can run on a computer from today, or 1970, or can be run by a guy with a pen and a sheet of a paper trained to mimic a general processor, at absurdly low rates of speed. That point is that you neeed a machine capable of running your program at a reasonable speed to be able to even develop the program. If a group of computer scientists from the 1970s were given detailed explanations of how modern processors and graphics cards work, and set out on the task of programming Doom 3 using the computers they had available in 1970, they were not be able to do it, even though theoretically, a pre-designed simulator of 32-bit processors and graphics cards and a copy of Doom 3 would run fine (just very, very slowly) on computers from the 1970s. A time traveller from the future could give us the answer and we could simulate the human brain very slowly, but the fact we can't run our software prohibits us from creating it. We can't run a simulation of the human mind in real-time, or at any reasonable speed. If computing power was somehow frozen at its current levels for the rest of time, we would never come up with human-equivalent AI. It is essential that we have increases in hardware performance to create human simulators, not just software.
Actually, a good number of science fiction writers are scientists. Gregory Benford, David Brin, and Alastair Reynolds are all currently employed as scientists, for example. Isaac Asimov was a scientist as well.
Furthermore, any novelist worth his/her salt does a lot of research to make sure they know what they're talking about. So when they get the future right, it's a well-informed guess, not so much a fluke.
I'll agree that they aren't necessarily brilliant geniuses, though.
I did some browsing and found a Wikipedia article that informs about this particular "singularity" term.
Also, here's some of Arthur C Clarke's predictions:
2002 Clean low-power fuel involving a new energy source, possibly based on cold fusion.
2003 The automobile industry is given five years to replace fossil fuels.
2004 First publicly admitted human clone.
2006 Last coal mine closed.
2009 A city in a third world country is devastated by an atomic bomb explosion.
2009 All nuclear weapons are destroyed.
2010 A new form of space-based energy is adopted.
2010 Despite protests against "big brother," ubiquitous monitoring eliminates many forms of criminal activity.
2011 Space flights become available for the public.
2013 Prince Harry flies in space.
2015 Complete control of matter at the atomic level is achieved.
2016 All existing currencies are abolished. A universal currency is adopted based on the "megawatt hour."
2017 Arthur C. Clarke, on his one hundredth birthday, is a guest on the space orbiter.
2019 There is a meteorite impact on Earth.
2020 Artificial Intelligence reaches human levels. There are now two intelligent species on Earth, one biological, and one nonbiological.
2021 The first human landing on Mars is achieved. There is an unpleasant surprise.
2023 Dinosaurs are cloned from fragments of DNA. A dinosaur zoo opens in Florida.
2025 Brain research leads to an understanding of all human senses. Full immersion virtual reality becomes available. The user puts on a metal helmet and is then able to enter "new universes."
2040 A universal replicator based on nanotechnology is now able to create any object from gourmet meals to diamonds. The only thing that has value is information.
2040 The concept of human "work" is phased out.
2061 Hunter gatherer societies are recreated.
2061 The return of Haley's comet is visited by humans.
2090 Large scale burning of fossil fuels is resumed to replace carbon dioxide.
2095 A true "space drive" is developed. The first humans are sent out to nearby star systems already visited by robots.
2100 History begins.
Beware: In C++, your friends can see your privates!
Of course, he got plenty of things wrong, just as did others. But that's not the point. I would actually argue that most of the stuff we enjoy today was at some point predicted in sci-fi. Not as a whole picture, but as particular ideas.