The dominant intelligence in the cosmos may be artificial
motherboard.vice.com
motherboard.vice.com
While I am inclined to agree with the overall premise here, this sort of thinking strikes me as a bit naive. Technological progress is not teleological, just as evolution is not teleological. (Technological development is not as random as evolution, certainly, but neither is it teleological.) There is no "best" way to develop technology, and there is certainly no predetermined, given course that technological development must inherently follow. Progress is heavily influenced by local circumstances, context, goals, and constraints. Even in studying human history, you encounter certain civilizations (the Mayans, for example, or the Greeks) who had remarkably "advanced" technologies in some domains, and the complete absence of expected technologies in other domains.
Instead, we should be talking about probabilities and correlations. The development of radio technology seems inherently linked to an understanding of the properties of electromagnetic radiation. We can probably assume that with some confidence. From there, what other technologies seem likely to be correlated to that knowledge, and with what degree of confidence? Etc.
"The Drake equation is a probabilistic argument used to estimate the number of active, communicative extraterrestrial civilizations in the Milky Way galaxy."
We are the only known intelligent species in the universe. So we model these exceptions after ourselves. Just like we search for earth like planets in the goldylock zone. Carbon and (possibly) silicon live forms are the only ones we know can happen.
The logic Shostak is using here begins with a sort of reverse-engineering of how humanity made its progress, followed by an implicit presumption that other advanced civilizations will follow the same path, until such time as they advance beyond us (by way of AI). I don't believe we can confidently say that technology has to progress along the path we walked.
On the other hand, many inventions and discoveries have been made simultaneously, and presumably this happens even more often than reported since it’s difficult to prove your invention is original once it is already known.
On the other hand, if you assume that strong AI can exist in a package at least as small as the human brain but without only a dependence on electricity instead of all the physical constraints mentioned above, your spacecraft could look much much smaller. If you give such a civilization 1000 years to iterate on the technology (an insignificant amount of time compared to the timescales involved in evolution), there's no good reason to think they wouldn't be able to shrink the computing technology (and therefore spacecraft technology) by several orders of magnitude. Consider the size difference and computing capacities of our earliest computers and smartwatches today--and we did that in less than 100 years.
The benefits are pretty obvious: it would require vastly fewer resources to travel to other stars, and any intelligence on board the spacecraft would be immortal (plus if it got "bored" on the way, it could power itself down and schedule itself to start up again once it got to where it was going). The stars seem out of reach to us because of the limitations of our lifespans and our bodies, but they would be entirely within the reach of AI. And from the perspective of conserving resources, there would be no reason to send something anywhere near the size of a human being.
I wonder if we're in some sort of temporary Golden Age at this point, mistakenly seeing the middle of a sigmoid curve and assuming perpetual growth. It wouldn't be the first time we underestimated strong AI. Still, I like it, the pervasive idea that nothing's impossible - it's just a matter of time.
If "strong" AI means software capable of learning any particular task a human could perform, we're so damn close that the research world is proceeding to cross the line as we speak. It's called "imitation learning" nowadays, and once there's a sufficiently general and accurate algorithmic framework with sufficient training data, it'll be everywhere.
If "strong" AI means software capable of learning many tasks a human could perform, generalizing them into a complete worldview, and taking volitional action based on that worldview to accomplish conceptually-specified goals... I'd call it one to two decades away. There are some major substantial problems remaining to research, but we do have solid foundations for it: uncertain reasoning in Turing-complete modeling domains lies at the intersection of several thriving research fields.
In terms of "how much science is left", I'd say our current point is roughly analogous to the physics community having spotted that the photoelectric effect can only be explained by discretized photons and starting to generalize quantum physics, but having most of it remaining to discover.
The reason nobody publicizes "AI" is because, from the perspective of a learning researcher or cognitive scientist, there basically is no such thing: there exists a large variety of cognitive tasks that can be modeled and pieced together in various ways, with varying degrees of accuracy when applied to real data-sets or compared to human performance.
Hopefully including what to do with the humans still dependent on the labor market that would immediately collapse on availability of that kind of AI.
In other words, when we can all live out our fantasies, what need do we have of reality?
Just because Greg Egan wrote a book about it doesn't mean it's a good idea.
wills it to turn blue
Nope, reality's still not mutable to the sheer extent that a solipsist virtual existence would be.
Honestly, if you want to not make everyone want to kill themselves rather than climb into your bizarre Matrix replica, you're going to have to enforce a few rules:
1) You can always get out, back to physical reality, or at least see physical reality.
2) Any other living things one may encounter are properly, independently alive. If they're people, then they're actually conscious with their own, independent minds.
3) Law of conservation of personal identity: people cannot be copied or altered, both to prevent fork-bombing and to preserve basic sanity.
Otherwise, the so-called virtual dreamlands are actually psychological torture chambers. People simply don't stay sane if you stick them in total isolation and then unhinge them from any causality independent of their own thoughts.
It's not that hard to turn your arm blue. Just paint it.
(Sorry for the overused argument, but I think it fits here. If we cannot ever decipher our constructed virtual reality from our current reality, why would we prefer one to another? Furthermore, there is no point to our current "reality", so how would a virtual world be different?)
Also any other version of visualization other than a perfect and indistinguishable (that makes it perfect in the definition of mirroring reality) constructed virtual reality is pointless to argue about. In your other reply, you mention issues with isolationism and being able to see the physical world. When the virtual reality is equivalent to the physical world, is this connection to physical reality necessary?
No, we know damn well that our current reality works based on fixed physical laws. You can't change something's color in reality by altering its texture-map, and you know that everything in reality which looks like an object is an object, while everything that looks like a person is a person.
>When the virtual reality is equivalent to the physical world, is this connection to physical reality necessary?
... Yes. In the exact same way that it is sometimes necessary to unplug your laptop, leave your cubicle, and go outdoors. For one thing, however much you might like your cubicle (read: "virtual reality"), it is ontologically dependent on outdoors (read: real life).
Or you could be walking along in virtual reality one day, enjoying the nice simulated weather from your upgraded Navier-Stokes package, when you neatly wink out of existence because a squirrel back in real life chewed the wrong cable.
> What I regret most is my uncritical treatment of the idea of allowing intelligent life to evolve in the Autoverse. Sure, this is a common science-fictional idea, but when I thought about it properly (some years after the book was published), I realised that anyone who actually did this would have to be utterly morally bankrupt. To get from micro-organisms to intelligent life this way would involve an immense amount of suffering, with billions of sentient creatures living, struggling and dying along the way. Yes, this happened to our own ancestors, but that doesn’t give us the right to inflict the same kind of suffering on anyone else.
> This is potentially an important issue in the real world. It might not be long before people are seriously trying to “evolve” artificial intelligence in their computers. Now, it’s one thing to use genetic algorithms to come up with various specialised programs that perform simple tasks, but to “breed”, assess, and kill millions of sentient programs would be an abomination. If the first AI was created that way, it would have every right to despise its creators.
His more recent story "Crystal Nights"[2] examines the same idea with more focus on the moral implications.
[1]: http://gregegan.customer.netspace.net.au/PERMUTATION/FAQ/FAQ...
Some problems are simple and random, some others are complex and deterministic and you also even get some problems that are are simple to state, deterministic, provably uncomputable, yet still understandable, such as the Wang tiling problem.
Roughly half the human population has below-average intelligence, but carries right on being human regardless.
Inventiveness is possibly more of a species marker than raw IQ.
We're not actually that great at being intelligent. But we're very good at using our intelligence to create experiences, invent new things, and pass both to offspring through indirect external memory structures.
>Are humans from 100,000 years ago less human than humans today, even if we have "improved" slightly due to natural selection?
It may be hard to find anyone who accepts that Paranthropus or Homo Erectus were just as human as modern Homo Sapiens.
Homo Erectus died out around 150kybp, which is close enough, I guess.
Unlike other animals where the key factor is morphology (horse -> giraffe) for humans it is the ability to operate as a group and ultimately as a society.
30kya sees hominids adapt to most areas of the world (northern asia, americas) and 10kya civizaltion starts. Although not the traditional biological definition, I would say that's the most salient factor.
Well, they would be intelligent in whatever way we mean it. But there's an issue of software -- 'human' implies quite a lot of it (for instance, things like the Westermarck effect, which I don't imagine AIs will have for quite a few reasons), and replicating that is pretty hard, so it's anyone's guess whether we do that or try to find a math-based software foundation.
Our future robots may be very good at passing information via some programming code... but will they love? Will they be sad? Will they read literature to empathize with the experience of other beings?
I would argue that "being human" is inextricably linked to the isolation of our existence (what David Foster Wallace described as being "marooned inside our own skulls", and why he argued that humans make literature to "give imaginative access to other selves"[1]) and the knowledge that our existence is finite. All humans -- from the cavemen all the way down to you reading this on you computer -- make life decisions knowing that our time on Earth is short. Do you share the same human experience if you know you will live forever?
[1] http://hudsonreview.com/2014/02/on-david-foster-wallaces-con...
It could very well be that we wouldn't even know that kind of intelligence exists - perhaps we're the ones crawling around in the ant farm?
That sentence gave me a flashback to my favorite literature reading robot - Orphu of Io (http://en.wikipedia.org/wiki/Ilium/Olympos)
Imho, this is basically a signal-response mechanism. You get some 'love' signal inputs and then you can process them into some response which reciprocates that love. It might turn into a hierarchical state system where a continual input of love signals can take a machine to a new state of 'deeply in love'.
If artificial is a designation meaning that it is created by humans, or another entity to serve a purpose, then what happens when that created AI has the ability to choose to create something itself? This "artificial intelligence" designation only works when a "creator" is trying to create an intelligence, then once it exists and propagates on its own, isn't it an "intelligence" of its own right?
My thought was, "a simulation versus what?" What makes our Universe a simulation or not a simulation? How does one define "simulation"?
It's kind of an artificial construct based on our acceptance of the current "reality" as somehow possessing this amorphous quality of "realness" in the first place.
In the end the "machines" will probably be better version of ourselves that will evolve artificially by DNA manipulation at faster rate than the current natural evolution.
I think something like this is more likely to be happening in the universe than our current view of self replicating evolving computers.
Actually I doubt this. Our brains and nerves are based on chemical processes which are very slow compared to electronics. Nerve impulses travel at about 60 MPH, versus a bit under the speed of light for fiber optics. That's around 40 billion times slower.
Also, our digital circuitry is approaching the size of individual molecules.
There are challenges with power and cooling, but they seem solvable.
"In the end the "machines" will probably be better version of ourselves that will evolve artificially by DNA manipulation at faster rate than the current natural evolution."
Engineering seems inherently far more efficient than evolution, artificial or otherwise. Even today we can build provably correct, arbitrarily large memories as long as we have hot-pluggable replacement parts. Parallel supercomputers also exhibit interesting traits of expandability, reliability and redundancy.
Many think that one of the first things that will be done with a first-generation "true AI" will be to have it engineer a second-gen AI. And so on...
"I think something like this is more likely to be happening in the universe than our current view of self replicating evolving computers."
Time will no doubt tell... :-)
No, it doesn't.
A lot of sucessfull creatures from the earth are just plain stupid, or maybe another kind of smart.
I wonder why nobody is creating for example a "sequoia" robot. A machine, designed with the main goal in mind not to move, talk or act like an animal, just to survive for 2000 years (and maybe store/deliver information in the meanwhile). This machine will not need to be very intelligent, will not need eyes or legs, not at all, just to be very robust and maybe self-reparating. Thats all.
A machine like this could be really useful for the human species. Instead we create chess players.
When we thing in a robot, everybody have in mind a machine able to speak and deliver company and smart punches and quotes. Well, this is perfect... for a hollywood film. In fact the aim should be to create a machine able to survive in a spaceship for "any time that we wish" or at least "a LOT of time", doing very simple things.
A machine don't needs to be intelligent; it needs to serve to an intelligent purpose.
So the answer to your hypothetical is mysticism as reality.
As a machine it was only capable of pure, cold logic with no emotion, but with its new-found sentience V'ger began to question its own existence. It asked the philosophical questions faced by so many lifeforms: "Is this all that I am? Is there nothing more?
While we can imagine a lot of things--like, for example, that everything interesting has already been invented, or that physics consists of cleaning up a few loose ends in classical mechanics, or that anyone going on about nuclear power is talking moonshine--the history of science has shown repeatedly and in great detail that what we imagine and what we can imagine doesn't amount to a hill of beans in this crazy world. The world is the way it is, and our imagination is pretty much an anti-tool for extrapolating from what we know into what we don't.
Utopians and distopians of all political and technological stripes tend to forget this. Which is OK so long as they keep their visions to themselves, or present them as the fictions they are (self-serving example: http://www.amazon.com/Darwins-Theorem-TJ-Radcliffe-ebook/dp/...). When they ask us to take them seriously they shouldn't be surprised that we respond skeptically, or with laughter.
It's fun to think about this stuff, but funny that anyone thinks their imaginings have a greater-than-epsilon chance of being true.
I'm hoping this doesn't come across as excessively mean-spirited, but I get tired of these high-flown fantasies being presented as if they had a non-negligible chance of being true. It promotes the idea that our unaided imagination is useful for prediction, and the data shows overwhelmingly that it is not, and that believing otherwise results in a considerable amount of preventable failure and human misery.
The most dangerous four words in any language are: "It just makes sense!"
While there is a limited value in such speculations, and they make a great basis for fiction, there is a tendency to take them far more seriously than they deserve. Science--the discipline of publicly testing ideas by systematic observation, controlled experiment and Bayesian inference--will tell us about the nature of the universe. Imagination is vital to coming up with ideas to test, but the silent universe--which isn't even mentioned in the article--is actual evidence that rather than super-intelligent AI, the dominant intelligence in the cosmos may (or may not) be us.
Maybe their manufacturing process is named "evolution"? :-) In that case they could just bring along some primitive life form. Who cares if it takes a few million years? And what if evolution isn't as random as we think? Or maybe the parts that are random are too trivial to care about, from the perspective of a few million years.
Settled on that it was created as opposed to evolving and it also wiped out its own creators.
I.e. the individual does not create in isolation. It took several generations of people to create to a point where we can even make computers.
So at least to me the whole divide between artificial and natural isn't really there.
Also, I find it funny that scientists always claim that the the future people will augment their intelligence with computers, to make them selves smarter. It's not a problem of capability, it's a problem of motivation. Have you ever heard someone say: "I wish I was smarter!" Maybe, but most of the time you hear people say "I wish I could have X, screw X, live in X." The simple fact is, most people couldn't care less than they already do about how intelligent they are, as long as they achieve the goals that make them happy.
In the future we will significantly increase the lifespan of humans, create very simplistic AI to do menial tasks for us, improve technology like 3D printing and molecular assembly, and then promptly have an economic collapse followed by a revolution that will lead to death of technology as we know it or utopia, hopefully the letter. I just don't see a place in that chain of events for a consciousness capable AI.
Case 1 for motivation: Governments want to be ahead of the governments of other countries, so they want smarter leaders. One way to accomplish that is via augmented humans, another is via supporting AI.
Case 2 for motivation: Businesses want more intelligent functioning and less employees (so they don't have to pay as much for wages). And they want the employees that they have to be brilliant at what they do. Economic forces are likely to fuel a substantial amount of progress with the development of more capable AI (just look at Google, Intel, Microsoft, Amazon...).
> Have you ever heard someone say: "I wish I was smarter!"
Yes, quite a bit actually. I certainly don't know anyone who wishes they were less intelligent.
Case 2: Though it's very true that all of the companies you have mentioned are working on AI systems, none of those systems need to be as intelligent as a human to be very good at the job they are intended for. Take Google Car for instance. Will Google some day soon have a car that will be possibly even safer at driving a car than a highly skilled human driver? YES! Does the AI in the car need to be even 1/2 as intelligent as my dog to do that task? No. Specialized AI is the future, general purpose AI is simply not necessary, and also, not really achievable any time soon.
> Yes, quite a bit actually. I certainly don't know anyone who wishes they were less intelligent.
Again, it's the motivations you have to pay attention to. Why do people want to be smarter? So they can have a better startup? So they can invent X, code X, build X? So, they don't want intelligence for intelligence sake, they want a tool. They don't want intelligence for it's own sake. Some people are smart, and as a result, very proud of the fact that they are smart. However, would they be as proud of their intelligence if they knew that at least part of it is augmented by an IBM chip. Again, there is no motivation to develop intelligence for intelligence sake. There are motivations to develop information syntheses "AI", but that would hardly supplant humans any time soon.
So they can know and understand everything possible in the Universe, make great contributions to humanity and the future, and live a pinnacle life full of fulfillment, riches and experiences few others attain. That's not to mention enduring fame throughout history.
As for artificial AI, that has the allure of mind uploading, and truly unimaginable power and experience including practical star travel.
Your views on this are weak.
However, it's a really good means to an end. Future people may not care about intelligence by itself, but they'll still augment their intelligence with computers because it's a really effective way to have fancy cars and nice houses and such.
This, of course, already happens. How many times have you heard programmers talk about how painful it is to program without an internet connection? It's a frequent refrain among the ones I know, limited only by the fact that it's pretty hard not to have an internet connection these days. We constantly use computers for intelligence augmentation already.
(I did RTFA and also agree with the author)
I don't want to say which books, though, because it's kind of a twist! No one spoil it!
Oyvaqfvtug ol Crgre Jnggf srngherf ulcrevagryyvtrag nyvraf gung ynpx pbafpvbhfarff. Pbafpvbhfarff vf cerfragrq nf n "oht" - va bgure jbeqf, fbzrguvat gung bpphef nybat jvgu vagryyvtrapr ohg vf abg fgevpgyl arprffnel sbe vg, naq gung cebivqrf pyrne qbjafvqrf.
Qvnfcben ol Tert Rtna srngherf nyvraf gung...jryy...vg pna'g ernyyl or fhzznevmrq rnfvyl, ohg fbzr penml fuvg unccraf (yvxr va znal bs uvf obbxf).