I would recommend anyone that is fascinated by this to read books from two universes; the Revelation Space universe, and the Culture novels. These feature two different looks at how AI & computers integrate or coexist with humans.
The Culture novels have both what is probably the best case for human-AI interaction in the Culture itself and what is possibly the worst case - societies using technology to implement virtual hells where uploads are tortured for eternity...
Edit: Are there any SF stories that give a more positive view of AI than the Culture novels?
I dunno, I'm no Culture series buff but from conversations with a friend who is I'm under the impression that To The Stars serial proposes a better job of having humans integrated into a unified Governance, though I suppose you could interpret the "just leave the AI types completely in charge" as the more AI-positive position.
Of course, the setting lets them cheat a little: it's basically HPMoR for magical girls. :b
Actually, I don't think the Culture is a case of "just leave the AI types completely in charge" - they are presented as having a lot of votes on things from whether you are allowed to build stuff (e.g. a network of cable cars) in a wilderness through to all-out-war. If a side looses a vote on something serious they regard it as natural that the society can split and nobody seems to regard this as unusual or undesirable.
One of the interesting things about the Culture novels is if you have god-like AIs (i.e. the Minds) how can a society control them? The answer the Culture gives is that the only real control is peer pressure e.g. the Grey Area:
Of course, any human in the Culture has to be comfortable with the idea that they are essentially well treated pets. Although pets that are free to leave at any time...
It's also mentioned that pan-humans can augment themselves intellectually to the point where they're at least in the same ball park as Minds. They're just no longer considered 'human' at that point.
I don't know if it's possible to have a more positive view of AI than the Culture books - the weakly godlike, (almost-)universally benevolent Minds are a bit too Mary Sue-ish and make Banks work hard to find interesting stories in the setting (which he does, and well, but he's forced to stick to the fringes of the Culture because most of it is too boringly perfect.)
I'm not sure that the Hells were particularly AI-heavy - the demons seemed to operate on a quite human level compared to the often-ineffable Minds.
Yes, one valid criticism is "one of the problems with the Culture novels as novels is that the central characters, the Minds, are too powerful and, to put it bluntly, too good."
What I find interesting in the culture series is how they talk about lesser AI being used by more powerful AI. I view this as being similar to what we have with Siri etc.
and the mysterious hallmark of Homo sapiens - our capacity to experience insights and apply intuition
AlphaGo demonstrated that it can find a better solution than players who spent their whole lives practicing intuition, which is limited by the brain's "capacity" and "computing power". If the problem is 'asking the right question', perhaps this can eventually also be brute-forced by machines faster.
That's not in any way the hallmark of Homo Sapiens. Animals of all sorts can improvise solutions in similar ways.
The real hallmark of Homo Sapiens is that culture and knowledge are externalised and persist outside individuals. So there's a social bootstrapping effect, because individual insights persist even after the humans who had the insights die.
Which means other individuals don't have to keep (literally) reinventing the wheel. Or quantum theory. Or technology in general.
If you take an individual human and don't expose him/her to culture or language, you get an animal with very ordinary and limited abilities.
This is one reason I think the current ML systems aren't necessarily looking in the right places. Learning is one thing, but systems teaching systems - or at least systems communicating insights via a common knowledge/abstraction framework/protocol might be something else again.
I think you are right in that animals of all sorts can improvise solutions in similar ways, most current ML systems aren't looking in the right places, and communication between systems is important.
Where I differ is that what is I believe animal intelligence (with human intelligence an example of to a higher degree) is actually incredibly powerful, and its that type of intelligence that we haven't quite been able to replicate yet. I think that combining highly parallel approaches like deep learning with virtually embodied AI will lead to animal-like general intelligence within a few years.
When that happens, when we have human level general artificial intelligence, the ability to perhaps instantly transfer granular knowledge between AIs, or at least, replicate full AI datasets, may be the biggest aspect that humans can't keep up with.
That is where things like DARPA brain computer interface research projects can help us to keep up.
> If you take an individual human and don't expose him/her to culture or language, you get an animal with very ordinary and limited abilities.
Can you suggest any research worth reading on this topic? I'd assume that a healthy human who was raised in isolation (if such a human could be considered healthy) would be seriously handicapped compared to properly raised humans but would still be significantly above all other animals in mental ability, but I'm really just guessing.
And you thought Silicon Valley was a free market Capitalist system. As the article notes, many of the big innovations have come out of Big Government spending of taxpayer dollars through agencies like DARPA. Yet most of the profits stay in private hands.
And you thought Silicon Valley was a free market Capitalist system
Who did? Certainly not people reading HN attentively, which gives plenty of coverage to technical developments by governmental institutions (and universities). DARPA alone has 2000+ stories about them.
Yet most of the profits stay in private hands.
Of course. Public institutions don't make a profit, by definition.
See the book The Entrepreneurial State by Mariana Mazzucato who covers this topic.
"Her book comprehensively debunks the myth of a lumbering, bureaucratic state versus a dynamic, innovative private sector. In a series of case studies—including IT, biotech and nanotech—Professor Mazzucato shows that the opposite is true: the private sector only finds the courage to invest after an entrepreneurial state has made the high-risk investments. In an intensely researched chapter, she reveals that every technology that makes the iPhone so ‘smart’ was government funded: the Internet, GPS, its touch-screen display and the voice-activated Siri. And in another chapter she argues that the green revolution is today is missing the kind of patient public sector financing, and de-financialized private sector, that got the IT revolution off the ground."
The title of this article is click-baity in its abuse of the vague, subjective word "consciousness."
"[blank] is already happening" is a trope, in wired's headlines, that gets reused frequently, to provoke exasperation. Wired probably advises writers with an internal style guide, since their tone has remained pretty consistent over the years.
The article is a lot of hype. There's a mix of several separate concepts brought into play, with anecdotal details, to produce an emotional effect.
Without explaining the differences between the peripheral nervous system and the central nervous system, ideas are presented to confuse such differentiation.
The article touches upon advanced ways to interface with the peripheral nervous system of the body, in particular the extremities, addressing paralysis and amputation.
The article is not about the brain itself, although witnesses within the article explain their personal experiences. It isn't about machines becoming alive. It isn't about replacing the mind with a computer.
It's mostly about advanced ways of restoring sensation and motor control, with computer systems that remap connections beween damaged nerves. The external stimulation of nervous tissue does not translate to a migration of human experience from a biological system to a machine.
I've noticed I started to avoid clicking links with titles I perceive as overly dramatic or emotional. I wish there was a browser extension providing community-sourced alternate titles and tl;drs for every news article on hover.
The specifics - molecular simulation feedback and prosthetic touch - are cool, but title is overblown. e.g. We've had REPLs and cochlear implants for decades.
This seems to talk about a weak form of human-machine integration.
Nevertheless, there are some fascinating examples. I remember how a Wired article talked about how advanced computing has made chess players like Magnus Carlsen so much better because he now fights against a computer that can overpower most if not all humans. It's made the marginal cost of practice so much lower as a result: you don't need the best grandmasters in the room to play against something that thinks like them.
As a result, similar to how better nutrition and health practices have augmented sports performance, I can see even this weak form of human-machine integration leading to incredible spurts in cognitive proficiency -- though perhaps it will be domain-limited for now.
29 comments
[ 3.0 ms ] story [ 112 ms ] threadEdit: Are there any SF stories that give a more positive view of AI than the Culture novels?
Of course, the setting lets them cheat a little: it's basically HPMoR for magical girls. :b
http://tvtropes.org/pmwiki/pmwiki.php/FanFic/ToTheStars
http://archiveofourown.org/works/777002/chapters/1461984
One of the interesting things about the Culture novels is if you have god-like AIs (i.e. the Minds) how can a society control them? The answer the Culture gives is that the only real control is peer pressure e.g. the Grey Area:
https://en.wikipedia.org/wiki/GCU_Grey_Area
Of course, any human in the Culture has to be comfortable with the idea that they are essentially well treated pets. Although pets that are free to leave at any time...
I'm not sure that the Hells were particularly AI-heavy - the demons seemed to operate on a quite human level compared to the often-ineffable Minds.
https://en.wikipedia.org/wiki/The_Culture#Overview
AlphaGo demonstrated that it can find a better solution than players who spent their whole lives practicing intuition, which is limited by the brain's "capacity" and "computing power". If the problem is 'asking the right question', perhaps this can eventually also be brute-forced by machines faster.
The real hallmark of Homo Sapiens is that culture and knowledge are externalised and persist outside individuals. So there's a social bootstrapping effect, because individual insights persist even after the humans who had the insights die.
Which means other individuals don't have to keep (literally) reinventing the wheel. Or quantum theory. Or technology in general.
If you take an individual human and don't expose him/her to culture or language, you get an animal with very ordinary and limited abilities.
This is one reason I think the current ML systems aren't necessarily looking in the right places. Learning is one thing, but systems teaching systems - or at least systems communicating insights via a common knowledge/abstraction framework/protocol might be something else again.
Homo Sapiens can do that, but I think it's more a consequence of our particularities that a cause.
I think the 'real hallmark' it's our capacity for abstraction: never mind how hard you try, you will never teach the Pythagorean theorem to a dog.
From the point of view or our current knowledge, there are not reasons for that being impossible to implement in software.
Also, why was the grandparent down-voted, please?
Where I differ is that what is I believe animal intelligence (with human intelligence an example of to a higher degree) is actually incredibly powerful, and its that type of intelligence that we haven't quite been able to replicate yet. I think that combining highly parallel approaches like deep learning with virtually embodied AI will lead to animal-like general intelligence within a few years.
When that happens, when we have human level general artificial intelligence, the ability to perhaps instantly transfer granular knowledge between AIs, or at least, replicate full AI datasets, may be the biggest aspect that humans can't keep up with.
That is where things like DARPA brain computer interface research projects can help us to keep up.
Can you suggest any research worth reading on this topic? I'd assume that a healthy human who was raised in isolation (if such a human could be considered healthy) would be seriously handicapped compared to properly raised humans but would still be significantly above all other animals in mental ability, but I'm really just guessing.
The problem, however, is that these were also inherently abuse situations and the victims may also have had other abnormalities.
Who did? Certainly not people reading HN attentively, which gives plenty of coverage to technical developments by governmental institutions (and universities). DARPA alone has 2000+ stories about them.
Yet most of the profits stay in private hands.
Of course. Public institutions don't make a profit, by definition.
"Her book comprehensively debunks the myth of a lumbering, bureaucratic state versus a dynamic, innovative private sector. In a series of case studies—including IT, biotech and nanotech—Professor Mazzucato shows that the opposite is true: the private sector only finds the courage to invest after an entrepreneurial state has made the high-risk investments. In an intensely researched chapter, she reveals that every technology that makes the iPhone so ‘smart’ was government funded: the Internet, GPS, its touch-screen display and the voice-activated Siri. And in another chapter she argues that the green revolution is today is missing the kind of patient public sector financing, and de-financialized private sector, that got the IT revolution off the ground."
http://marianamazzucato.com/projects/the-entrepreneurial-sta...
"[blank] is already happening" is a trope, in wired's headlines, that gets reused frequently, to provoke exasperation. Wired probably advises writers with an internal style guide, since their tone has remained pretty consistent over the years.
The article is a lot of hype. There's a mix of several separate concepts brought into play, with anecdotal details, to produce an emotional effect.
Without explaining the differences between the peripheral nervous system and the central nervous system, ideas are presented to confuse such differentiation.
https://en.wikipedia.org/wiki/Peripheral_nervous_system
The article touches upon advanced ways to interface with the peripheral nervous system of the body, in particular the extremities, addressing paralysis and amputation.
The article is not about the brain itself, although witnesses within the article explain their personal experiences. It isn't about machines becoming alive. It isn't about replacing the mind with a computer.
It's mostly about advanced ways of restoring sensation and motor control, with computer systems that remap connections beween damaged nerves. The external stimulation of nervous tissue does not translate to a migration of human experience from a biological system to a machine.
I've noticed I started to avoid clicking links with titles I perceive as overly dramatic or emotional. I wish there was a browser extension providing community-sourced alternate titles and tl;drs for every news article on hover.
... Absolutely nothing in the text substantiates any of the very strong claims.
Nevertheless, there are some fascinating examples. I remember how a Wired article talked about how advanced computing has made chess players like Magnus Carlsen so much better because he now fights against a computer that can overpower most if not all humans. It's made the marginal cost of practice so much lower as a result: you don't need the best grandmasters in the room to play against something that thinks like them.
As a result, similar to how better nutrition and health practices have augmented sports performance, I can see even this weak form of human-machine integration leading to incredible spurts in cognitive proficiency -- though perhaps it will be domain-limited for now.