Ugh, I know this is par for course, but these kind of articles really irritate me. Especially passages like these:
> People with connected smartphones or tablets anywhere in the world have access to many (if not most) of the same communication resources and information that we do while sitting in our offices at MIT. They can search the Web and browse Wikipedia. They can follow online courses, some of them taught by the best in the academic world. They can share their insights on blogs, Facebook, Twitter, and many other services, most of which are free. They can even conduct sophisticated data analyses using cloud resources such as Amazon Web Services and R, an open source application for statistics.
A. What does this have to do with artificial intelligence?
B. Access to technology does not equal access to ability to use that technology. I would argue that the problem isn't that people don't have access to powerful statistical processing and analytical software, it's that they don't know what to do with it, or with the plenty powerful software that has been available to them in the past decade. The parallel construction of "Everyone can now share stuff on Facebook" with "Everyone can now do statistical analyses via R and Hadoop!" is a gross error of characterization by the OP.
Some academic types have no clue how far behind the rest of humanity is in terms of intellectual and technological readiness to use such tools. It could be an honest mistake and not a deliberate lie.
Compression doesn't seem that smart on the level of 'gzip'. But one can also look at what the brain is doing when learning (and classifying stuff into conceptual clusters) as also being compression. Given the enormous theoretical 'capacity' of the brain (however defined), the compression may not be required for space reasons - but it potentially serves an interesting function in terms of understanding.
What scares me the most is that by the time we have the technology to bring this artificial intelligence very close to our limbic system and subconscious mind, our computers (smartphones, desktops, tablets and other) will be so totally and thoroughly monitored, throttled, spied on, back doored, and censored, that we will be presented with the choice:
1. Refrain from getting enhanced, and suffer the consequences of limited ability and limited existence.
2. Get enhanced, and start the slow march where people aren't individuals anymore, we are drones in a collection of drones, with a queen.
But hey, maybe that's the best thing ever to happen to our species, to become half and half. Half of the humans become like a hive of bees or colony of ants, and the other half sees if separate and individual free agents is able to perform better. Then the wars that come later will see which one natural selection selects.
Hives and colonies where the drones don't contribute to the direction are better at staying alive and brute forcing any goal, mowing down an opponent through violence, keeping the colony and hive alive against the individual interests of the drones.
Whereas a collection of cooperating individuals are better at choosing the best opportunities for the individuals and advancing the cause of the interests of the drones. Cooperating individuals cause chaos, infighting and huge inefficiency in things that should be very easy, however, in doing so, each individual becomes like a mini hive/colony in and of itself.
We are are the cusp of a glorious age, we may live to see the creation of a new species of human which are to humans as humans are to monkeys. All made possible through using technology to enhance all of our senses, and to provide additional resources to our own thoughts.
I just might join that colony of bees. In the future, I think militaries will take this route. When you join the military and are deployed, you give up your individuality, your movements and thoughts are no longer your own.
I don't think that will ever happen. Anything more than primitive enhancement is extremely complicated and difficult. We will likely get strong AI and have a singularity first. Then AI will either kill us all or solve all our problems for us.
Why would it leave? The Earth is full of resources it can use and humanity would be a potential threat if left alone (we could build more AI at least.)
It might not need all the resources, and it might decide that the chance of humanity being a threat is outweighed by what it would take to solve the problem.
You're giving humanity far too much credit here--we ourselves basically ignore the animal kingdom from which we came unless it gets in our way.
It depends how strong the AI is but destroying humanity might be as simple as unleashing a superlethal virus, or a swarm of self-replicating nanobots (yes it sounds like scifi but we are talking about minds possibly thousands of times more intelligent than humans.) Then you have all of the Earth's resources and no threat from humanity whatsoever.
We are the technology. Resistance is futile. Your technological and cultural distinctiveness will be added to our own. Your culture will adapt to service us. We are the technology, resistance is futile.
I'm starting to think that an intelligent species with the social hierarchy that we have, can only go to a certain level of advancement.
Then it becomes impossible to move further, without changing the social schema. Possibly into a hive form, where everything is easier to manage.
From an engineering point of view, it's the perfect choice: easy to manage, easy to enhance, divide labour and what not. From a social point of view, yikes.
Is there a branch of sociology or something that study the effects of different social schemas?
But then we might not have come here without our social hierarchy, exactly because it is harder to manage. It makes people want to outsmart each other. (This is actually a hypothesis in evolution theory.)
It's arriving right now. I can speak in plain English to my smartphone (or any other language really, it knows more human languages than I do), cars drive themselves, the world's billionaires are now focused on rapid AI development, etc. There are so many things different this time around that you shouldn't be so pessimistic. "It's different this time" (it actually is though)!
The difference to me is that the dolls capable of playing music were made up of a small set of parts compared to 15 million lines of code for Wolfram Alpha. We are way more sophisticated today.
These are indeed improvements, but I don't think they qualify as true AI. Software is still nowhere near being able to pass the Turing Test. We still don't have any idea how the human mind works, or how to emulate one digitally.
Isn't it a bit of a disservice to the hardware side to say that just Software has improved? Think back 10 years, 2004...would the Nest have been possible? How about Google's self driving car? I think these things go hand in hand.
Somewhat, but keep in mind that for a hard problem (and AI is a very, very "hard" problem, software is going to be the bottleneck to thinking machines.
I feel like this article is just another aspect of the bubble. We have people on the sidelines-- intelligent people, but unaware of what the "tech" industry is actually doing-- cheering on something they hardly understand.
Look, Google is a closed-allocation company. So, I believe, are Apple and Amazon and probably Facebook. Socially, we're stuck in the 1980s. If people at the 99th percentile of technical literacy are still stuck mindlessly churning tickets in closed-allocation environments-- note that basic research funding has fallen off a cliff in the past 30 years-- and raising VC is still more about playing a feudal reputation economy than anything remotely approaching excellence, then what are most of the world's 7.1 billion people going to be able to contribute to this brave new world? Nothing. They'll have no leverage, nor the resources to learn what they need to learn because they'll have to spend all their time on basic survival. The only thing that seems to await the majority, barring some unforeseen social or economic right angle (the long-term odds of which are quite high) is joblessness.
We can't even deliver basic healthcare to middle-class Americans anymore, and here we have the gall to say that humanity's approaching some technotopia?
I agree with the healthcare sentiment. Obamacare is better than bankruptcy though. A $12k max out of pocket vs say $150k+. At least one could attempt to pay that off and maybe get some kind of Federal forgiveness vs straight out bankruptcy.
AI is limited more by language than anything else. NLP which I do a lot of work on at http://plexinlp.com goes a long way, but the computer doesn't have the ability to get metaphors, or sarcasm, or have any sort of "BS" detection.
CoreNLP (which isn't as good as our NLP but is the "standard" only gets the parts of speech right 95% of the time, by words, but about 75% of the time by sentence. Imagine you were a waiter and you only understood 75% of sentences. You'd not last too long.
But somehow we expect computers which not only are limited to 75% of sentences being "tagged" correctly, but then not having any idea about the concepts behind them.
AI as we see it today is not so much "AI" as it is Computer Driven Bucketing. Build a formula that gets as many things in to pre-sorted buckets as possible. For me it isn't AI until the computer can make leaps of its own. Create a new bucket and explain why it was needed, or create a new solution that wasn't proposed by the programmer.
I am working hard to bring all of that to fruition, but honestly we are a long ways off.
Here are some milestones I think we will need to hit before we get to AI.
Software powerful enough to convert Wikipedia (or any knowledgebase) to structured Data automatically.
Query system that takes any user input and converts it to something resembling a Database query (This is what I'm working on)
NPC's for Games/MMO's etc that pass a turring test with in the limits of their Virtual world. (Basically they don't have to pass for real, just as real as the DnD character a human would play)
These are really "simple" milestones that are the step before we see true AI start to work.
But why do we need to judge ai by our language? if were dealing with data and images, machine learning does do impressive things,including things that if were done by researchers would have certainly gotten them a place inside respectable academic journals.
For example ,a machine learning algorithm devised a new internet protocol that doubles the network capacity over best known methods.Another famous machine learning system understood own it's own how to classify images what belongs to the bucket of "cat" and many other object types.
Why isn't this enough to be considered intelligent?
Because those aren't Intelligent, they are just "handy".
Sorting doesn't require intelligence. You have X number of attributes and using them you do your best to get as many things in to "Like" buckets as possible.
Discovering new attributes is the start of intelligence. Discovering new buckets could be intelligence.
If the machine gets 100k pictures and can sort them in to Cat vs No Cat, and then suggests a new set of classification "Indoors vs outdoors" using attributes the developer didn't pre-identify that would be intelligence.
Google's famous experiment did something similar: it offered the machine plenty of content(from youtube) and the machine on it's own discovered the category "cat", and others.
You don't seem to want to have a discussion. So this is my last comment.
It was given all the possible traits. It didn't find a new category it had a collection of things with known traits. Google knew it would be able to find cats using the traits it was given. That's not intelligence on the computer's side.
If I tell you that there are things with lungs, things with gills, things with legs and things with arms, things with hair, I don't have to tell you there are mermaids and salamanders, you will figure it out when not everything fits in the bucket of Cats and goldfish.
As much as I'm looking forward to all the shiny tech I can't help but feel anxious about the implications for humanity as a whole.
e.g. Right now the poorest of the poor can still resort to back breaking repetitive work to survive. If the tech keeps going then soon these people will not be able to compete...at all.
Its a recipe for inequality magnified 100x & I don't see the social fabric surviving that.
Personally I oppose the idea - kinda. To be more precise I would prefer a system that stacks the odds in their favour heavily rather than "free money". e.g. Give them their cash...but make them guide school kids across a dangerous road in return. i.e. Extract some (extra) social benefit from it.
Makes little financial sense either way, but I'd much rather have them do the school kids thing & collect cash than sleep & collect cash.
Some people talk about "Guaranteed Minimum Income" as a conditional income guarantee, as opposed to an unconditional "Basic Income"
I don't feel strongly (either approach would be an improvement over the status quo), but I suppose I mildly agree with you. I think human beings need to have some Work to feel like their lives have meaning. Even if we made GMI conditional on 10 hours a week of charity volunteering, that could have positive mental health outcomes.
Do you have any citations that mandatory charity work are beneficial for the psychological condition? Also are there data about negative or positive externalities of forced charity work?
>Do you have any citations that mandatory charity work are beneficial for the psychological condition?
You raise a valid point. At first I thought "don't be stupid...obviously its beneficial". Looking at that word "mandatory" in your question though I'm no longer confident in my initial instinct....
In our hypothetical scenario, it's not "mandatory"--it's freely available. If you already have some other highly paid job, great. If not, there's always the option of doing the volunteer work in exchange for the minimum income from the government.
There would probably also need to be some kind of exemption for people with physical or mental disabilities.
Exemptions and regulations add to the administrative overheads. Monitoring and enforcing quality of volunteer work is a large overhead, too. The efficiency of basic income lies in its extreme simplicity.
However, the details of basic income-like schemes may be irrelevant. Basic income and its mutations are possible when most household products are basically done by robots. It is possible basic income will grow at 25-50% YoY rate and that growth indicator is only thing that matters.
Like I said, I'm not really invested in basic income vs guaranteed minimum income.
You could have a volunteer work "requirement" that was only enforced by social convention. If it was seen as dishonest/lazy to take the income without giving back to your community in any way, that would probably spur some social good without the inefficiency of a heavy-handed enforcement mechanism.
> The efficiency of basic income lies in its extreme simplicity.
Fair enough. I find it difficult to accept a system that doesn't grant a powerful advantage to a person with a significant physical/mental disability though. Equal basic income for all does not seem fair in that context.
Also...this is more me thinking out loud than attacking anyone.
I get what you are saying, but what about if we optimize not for equality but for happiness. Redistribution of value is a zero-sum game, while non-material value as happiness is not.
I was thinking how equality movemen failed to produced social change in short time as world is extremely unbalanced. I think post-scarcity economy will bring new solutions and new social contracts.
>what about if we optimize not for equality but for happiness.
Again, you make a valid point. Let me stop the debate right here though. We're venturing into philosophical territory here...the kind more suited to 5000 word essays than HN posts.
One thing we can debate however:
>post-scarcity economy
I find this notion decidedly problematic. We've been optimizing the sht out of production for decades and yet the average family is battling to make ends meet more than ever. "post-scarcity economy"...I don't buy it...not even close.
I get where you are coming from and I agree that the current track record doesn't allow for such predictions. However, I am linking the advent of post-scarcity to the widespread adoption of some transhumanism values. Maybe that'll work. Or we'll be consumed by grey goo. The future is too bright to see it clearly.
On the somewhat unrelated note my favorite flavor of transhumanism is abolitionism that aims to remove suffering as much as possible.
I suspect we're on the same (idealistic) philosophical wave length here, I just can't link my current (personal) projections of the future with what you describe.
You speak of "remove suffering" whilst I speak of inequality increased a 100 fold. You are clearly a very reasonable person, but I don't see a "meeting of minds" happening here unless I misunderstood you dramatically - there is no overcoming that gap. A 100 fold increase in inequality destroys everything in its path.
Basic income wouldn't take work away. It would assure that not working when there isn't work to do wouldn't leave people impoverished or lead to an economic collapse.
It would allow people to participate in the economy and pursue productivity or creative pursuits without worrying about starving.
I doubt it's coming, but I don't see its downside in a world where per capita productivity is trivial.
>need to have some Work to feel like their lives have meaning.
Agreed.
Vaguely relevant story time for 1st world crowd: In my country (South Africa) it is illegal for the public to refuel their own cars. There is dedicated staff that does nothing else all day - essentially behave like restaurant waiters but for petrol stations.
Realistically the law is only in place for job creation purposes - it is by most measures unnecessary. And yet I've come to appreciate it: I don't need to leave my car. The people handling flammable liquids do this all day long & received training - much better than random McDonalds munching idiots. Credit card speedpoint gets brought to my car window, windscreen washed, tyre pressure checked etc. I end up (voluntarily) tipping 0.5 USD - a drop in the ocean. I have a million problems so a guy making my life easier there & smiling whilst doing so...I'm totally ok with carrying him on my tax/fuel bill.
Thats the point - it could be done "easier" via self-help or automated but the law requires job creation. Us privileged people should seriously consider accepting that while it does not make economic sense, it does work from a convenience & social point of view. Both of those are powerful drivers.
> It would be better just to pay people for nothing rather than requiring them to do useless work.
You miss my point entirely. It is not useless work. Sure its not economically efficient, but on balance I'd rather have people streamline the refueling process than sit in front of a TV and collect a paycheque.
And no it is not one. There is a big difference between "economically efficient" and "wasting their times on a unnecessary job", especially in the context of GMI.
>I'd rather have people streamline the refueling process than sit in front of a TV and collect a paycheque.
You make it look that these two are the only alternatives when they are not (e.g. people could also be studying or actively searching for a job). Pretty much the definition of false dichotomy isn't it?
Why? They could get a hobby or another job. Some socialists argue that employment is morally wrong because it's the equivalent of slavery. I never bought that because we currently need people to work in order for the economy to function. But if working is really unnecessary and force people to work anyways, well what is that?
>Sounds like something that could be easily automated.
Yes and no. You underestimate the experience imo. e.g. Today I had two people dressed in red shirts (for valentines day) focusing on me, my car & what I wanted. Yes you can automate the physical refueling, but can you automate the conversation about the Tshirts specifically bought for Valetines day?
For a minor cost I can:
> Create a job
> Interact with a person in a different class
> Ease my daily burden
> Improve safety
> Assist the local economy (micro scale)
> Improve social balances
As I said...I tip 0.5USD at petrol stations and I could totally get away with zero.
Yes the mechanical act could be automated but that is entirely besides the point.
In the United States, Oregon and New Jersey also require an attendant pump gas for you. Supposedly gas stations have lower insurance premiums as a result.
That is not what I meant. You speak of idealistic implementation...its not all that far fetched. Refer to my comment here for a real life working implementation:
Perhaps. I think you judge them unjustly though - it is certainly prone to bureaucrats sinking the scheme with BS, but I don't think it is in itself prone to sinking.
As for the TSA comment - my condolences to all Americans. I don't feel its a representative example of the current debate though.
My big objection to this is that you have to pay a bunch of government bureaucrats to sit around and say what does and doesn't count. I don't think that's a worthwhile use of people's time.
The authors of the article actually wrote a book on this very subject (mentioned at the bottom of the article). They actually devote a lot of attention to exactly that problem, which they refer to as the bounty (gains because of new technology) and the spread (the inequality you mention).
Thanks - I didn't see that. Might actually buy that book.
As I said this is something that makes me deeply uneasy. I'm pretty relaxed about most "serious" issues...e.g. Climate change - I feel "it'll be fine & we'll make it work somehow". Looking at this problem I can't help but feel it will not be OK. And thats coming from a guy in South Africa - a society that is barely hanging on under the strain of inequality. SA is on the very edge of what the social fabric can withstand in terms of inequality & AI will cause inequality that is much worse.
Plainly the best thing to do is hurry the development of machine learning algorithms for active environments, also known as AI, in order to minimize the gap of time between most of the world's people becoming unemployable and most of the world's people being made of atoms the AI can use for something else.
The thing that's so aggravating about this article is what is so aggravating about AI: the arrogant comparison of software systems to true intelligence. The examples cited by the article are emphatically not "intelligence" as we understand it: Siri isn't, Watson isn't, Google's self-driving cars aren't. They represent algorithms (clever, in many cases) executed by a stored-program computer -- or many. They take advantage of what computers are really good at: the ability to perfectly execute endless streams of simple instructions with astonishing speed; the ability to access vast quantities of data with zero error; and the ability to reliably transmit that data to one another. These innovations also reflect the fundamental brittleness of the machines that implement them: the machines can only do exactly what they're told -- and if they're told to do the wrong thing, they will gladly do it. The intelligence, then, is purely in the humans that are programming and engineering the thing -- not the thing itself. I understand that this distinction is entirely lost on those who have no understanding of what's behind the curtain -- but we do a broader disservice when we pretend that our adept sleight of hand is, in fact, magic.
And my apologies for just speaking the truths that many here already know; over my decades in computer science, I have tried to hold my tongue whenever the AI cultists scream at passers-by that the singularity is near -- but every now and again I must scream back for my own sanity.
It doesn't seem like you actually read the article, but are just commenting on the title. They're not saying any of these innovations are true intelligence. What they're saying is that computers (and robots) are starting to do things that until recently were thought to be solely the domain of humans. They're saying that we're at the start of a new industrial revolution (a second machine age) which will lead to enormous increases in productivity.
This article focuses on the positive effects. In their book they actually spend quite a lot of time discussing potential negative effects as well, such as unemployment. You'll be glad to hear that they agree with you that jobs like programming and engineering will be ever more important in that world.
As much as I might like to deny it, I did, in fact, read the article:
Because the exponential, digital, and recombinant powers of the second machine age have made it possible for humanity to create two of the most important one-time events in our history: the emergence of real, useful artificial intelligence (AI) and the connection of most of the people on the planet via a common digital network.
"The emergence of real, useful artificial intelligence" is what I dispute; if the article chose to describe this as the "emergence of real, useful, algorithms and systems to address meaningful problems" then, fine. But of course, that isn't news, either -- it's merely the story of humanity in a sentence: it is us using our bipedalism and our oversized brains to do what we have always done to survive and thrive.
Once you understand the algorithm, suddenly it doesn't seem mysterious and then it can't be "true" AI. I'll agree that advancements in machine learning aren't necessarily leading to strong AI, but they are still incredible advancements. Many problems in the field of AI have been solved, and then suddenly dismissed as not being "AI".
Some of us believe that general intelligence is possible in (complex) software on a (powerful) stored-program computer. As in, we don't know how to write it yet but it doesn't require weird quantum effects. The processors are just doing the very simple operations they are told to, very fast, but the intelligence arises from the complexity of the system.
However I quite agree that the cheerleaders for AI frequently do us all a disservice by exaggerating the successes, and above all by exaggerating how close we may be to a breakthrough.
It depends a bit on what you mean by "intelligence as we understand it". As I see it animals have varying levels of intelligence - even amongst flies buzzing around the room I notice some of them just blindly crash into things but some of the larger species understand the layout of the room including windows and fly about without crashing. At a higher level Orangutans have been known to watch their keeper lock the door and then try to pick the lock. I don't think you can say only humans who can say play chess are intelligent and the rest doesn't count. On this scale computer systems are at around the smart fly level and improving. As regards the the argument that AI is just human developed algorithms, you and I are probably just evolution developed networks of neurons. How does that prove whether either system is smart or not?
AI is marching forward relentlessly. It may or may not be as close as most people think, but were are still getting there within a generation or two at most. Meanwhile almost nothing is being done to solve the friendliness problem. Since it seems to be much more complicated than building normal AI it may never be solved. There might not be a solution - our morality and values are just a bunch of messy evolutionary heuristics, and abstract concepts are hard to define concretely (eg "a human is a bunch of atoms in this specific configuration, but not this specific configuration. It's ok to kill one but not the other.")
The effects of AI are an extension of human will. We will succeed or fall with or without AI. AI will only be a factor or artifact in this process, not its determinant.
AI brings the possibility of creating minds literally millions of times more intelligent than humans. That kind of power is unimaginable to beings in a world where the dumbest village idiot still shared 99% of his brain structure in common with the greatest genius. An AI will either solve all our problems pretty much instantly, or crush us like ants.
In terms of general intelligence, that's probably true, but in terms of accomplishing tasks that are useful to humans, AI is far ahead of rats. There are many useful tasks that I can get an AI to do for, but practically none that I can get a rat to do for me.
111 comments
[ 105 ms ] story [ 4604 ms ] thread> People with connected smartphones or tablets anywhere in the world have access to many (if not most) of the same communication resources and information that we do while sitting in our offices at MIT. They can search the Web and browse Wikipedia. They can follow online courses, some of them taught by the best in the academic world. They can share their insights on blogs, Facebook, Twitter, and many other services, most of which are free. They can even conduct sophisticated data analyses using cloud resources such as Amazon Web Services and R, an open source application for statistics.
A. What does this have to do with artificial intelligence?
B. Access to technology does not equal access to ability to use that technology. I would argue that the problem isn't that people don't have access to powerful statistical processing and analytical software, it's that they don't know what to do with it, or with the plenty powerful software that has been available to them in the past decade. The parallel construction of "Everyone can now share stuff on Facebook" with "Everyone can now do statistical analyses via R and Hadoop!" is a gross error of characterization by the OP.
B. Sure.
Facebook has to crop photos in an optimal way - they do a pretty good job of that.
Either of those feel closer to AI than compression, which is sometimes given as another example of an AI challenge.
If I give you a pile of text, what compression ratios would you achieve by hand?
1. Refrain from getting enhanced, and suffer the consequences of limited ability and limited existence. 2. Get enhanced, and start the slow march where people aren't individuals anymore, we are drones in a collection of drones, with a queen.
But hey, maybe that's the best thing ever to happen to our species, to become half and half. Half of the humans become like a hive of bees or colony of ants, and the other half sees if separate and individual free agents is able to perform better. Then the wars that come later will see which one natural selection selects.
Whereas a collection of cooperating individuals are better at choosing the best opportunities for the individuals and advancing the cause of the interests of the drones. Cooperating individuals cause chaos, infighting and huge inefficiency in things that should be very easy, however, in doing so, each individual becomes like a mini hive/colony in and of itself.
We are are the cusp of a glorious age, we may live to see the creation of a new species of human which are to humans as humans are to monkeys. All made possible through using technology to enhance all of our senses, and to provide additional resources to our own thoughts.
I just might join that colony of bees. In the future, I think militaries will take this route. When you join the military and are deployed, you give up your individuality, your movements and thoughts are no longer your own.
Why, when killbots are much cheaper?
You're giving humanity far too much credit here--we ourselves basically ignore the animal kingdom from which we came unless it gets in our way.
Isn't this the best solution long term? Makes sense to me.
Then it becomes impossible to move further, without changing the social schema. Possibly into a hive form, where everything is easier to manage.
From an engineering point of view, it's the perfect choice: easy to manage, easy to enhance, divide labour and what not. From a social point of view, yikes.
Is there a branch of sociology or something that study the effects of different social schemas?
Look, Google is a closed-allocation company. So, I believe, are Apple and Amazon and probably Facebook. Socially, we're stuck in the 1980s. If people at the 99th percentile of technical literacy are still stuck mindlessly churning tickets in closed-allocation environments-- note that basic research funding has fallen off a cliff in the past 30 years-- and raising VC is still more about playing a feudal reputation economy than anything remotely approaching excellence, then what are most of the world's 7.1 billion people going to be able to contribute to this brave new world? Nothing. They'll have no leverage, nor the resources to learn what they need to learn because they'll have to spend all their time on basic survival. The only thing that seems to await the majority, barring some unforeseen social or economic right angle (the long-term odds of which are quite high) is joblessness.
We can't even deliver basic healthcare to middle-class Americans anymore, and here we have the gall to say that humanity's approaching some technotopia?
"If the world doesn't operate at what I deem to be just the right level of openness, we're all doomed."
CoreNLP (which isn't as good as our NLP but is the "standard" only gets the parts of speech right 95% of the time, by words, but about 75% of the time by sentence. Imagine you were a waiter and you only understood 75% of sentences. You'd not last too long.
But somehow we expect computers which not only are limited to 75% of sentences being "tagged" correctly, but then not having any idea about the concepts behind them.
AI as we see it today is not so much "AI" as it is Computer Driven Bucketing. Build a formula that gets as many things in to pre-sorted buckets as possible. For me it isn't AI until the computer can make leaps of its own. Create a new bucket and explain why it was needed, or create a new solution that wasn't proposed by the programmer.
I am working hard to bring all of that to fruition, but honestly we are a long ways off.
Here are some milestones I think we will need to hit before we get to AI.
Software powerful enough to convert Wikipedia (or any knowledgebase) to structured Data automatically.
Query system that takes any user input and converts it to something resembling a Database query (This is what I'm working on)
NPC's for Games/MMO's etc that pass a turring test with in the limits of their Virtual world. (Basically they don't have to pass for real, just as real as the DnD character a human would play)
These are really "simple" milestones that are the step before we see true AI start to work.
For example ,a machine learning algorithm devised a new internet protocol that doubles the network capacity over best known methods.Another famous machine learning system understood own it's own how to classify images what belongs to the bucket of "cat" and many other object types.
Why isn't this enough to be considered intelligent?
Sorting doesn't require intelligence. You have X number of attributes and using them you do your best to get as many things in to "Like" buckets as possible.
Discovering new attributes is the start of intelligence. Discovering new buckets could be intelligence.
If the machine gets 100k pictures and can sort them in to Cat vs No Cat, and then suggests a new set of classification "Indoors vs outdoors" using attributes the developer didn't pre-identify that would be intelligence.
It was given all the possible traits. It didn't find a new category it had a collection of things with known traits. Google knew it would be able to find cats using the traits it was given. That's not intelligence on the computer's side.
If I tell you that there are things with lungs, things with gills, things with legs and things with arms, things with hair, I don't have to tell you there are mermaids and salamanders, you will figure it out when not everything fits in the bucket of Cats and goldfish.
e.g. Right now the poorest of the poor can still resort to back breaking repetitive work to survive. If the tech keeps going then soon these people will not be able to compete...at all.
Its a recipe for inequality magnified 100x & I don't see the social fabric surviving that.
We need to start thinking about basic income sooner rather than later: http://www.bostonglobe.com/ideas/2014/02/09/should-governmen...
Personally I oppose the idea - kinda. To be more precise I would prefer a system that stacks the odds in their favour heavily rather than "free money". e.g. Give them their cash...but make them guide school kids across a dangerous road in return. i.e. Extract some (extra) social benefit from it.
Makes little financial sense either way, but I'd much rather have them do the school kids thing & collect cash than sleep & collect cash.
I don't feel strongly (either approach would be an improvement over the status quo), but I suppose I mildly agree with you. I think human beings need to have some Work to feel like their lives have meaning. Even if we made GMI conditional on 10 hours a week of charity volunteering, that could have positive mental health outcomes.
You raise a valid point. At first I thought "don't be stupid...obviously its beneficial". Looking at that word "mandatory" in your question though I'm no longer confident in my initial instinct....
There would probably also need to be some kind of exemption for people with physical or mental disabilities.
However, the details of basic income-like schemes may be irrelevant. Basic income and its mutations are possible when most household products are basically done by robots. It is possible basic income will grow at 25-50% YoY rate and that growth indicator is only thing that matters.
You could have a volunteer work "requirement" that was only enforced by social convention. If it was seen as dishonest/lazy to take the income without giving back to your community in any way, that would probably spur some social good without the inefficiency of a heavy-handed enforcement mechanism.
Fair enough. I find it difficult to accept a system that doesn't grant a powerful advantage to a person with a significant physical/mental disability though. Equal basic income for all does not seem fair in that context.
Also...this is more me thinking out loud than attacking anyone.
I was thinking how equality movemen failed to produced social change in short time as world is extremely unbalanced. I think post-scarcity economy will bring new solutions and new social contracts.
Again, you make a valid point. Let me stop the debate right here though. We're venturing into philosophical territory here...the kind more suited to 5000 word essays than HN posts.
One thing we can debate however:
>post-scarcity economy
I find this notion decidedly problematic. We've been optimizing the sht out of production for decades and yet the average family is battling to make ends meet more than ever. "post-scarcity economy"...I don't buy it...not even close.
I get where you are coming from and I agree that the current track record doesn't allow for such predictions. However, I am linking the advent of post-scarcity to the widespread adoption of some transhumanism values. Maybe that'll work. Or we'll be consumed by grey goo. The future is too bright to see it clearly.
On the somewhat unrelated note my favorite flavor of transhumanism is abolitionism that aims to remove suffering as much as possible.
You speak of "remove suffering" whilst I speak of inequality increased a 100 fold. You are clearly a very reasonable person, but I don't see a "meeting of minds" happening here unless I misunderstood you dramatically - there is no overcoming that gap. A 100 fold increase in inequality destroys everything in its path.
But in any case, compulsory community service as an alternative to jail time has shown promising results: http://www.bbc.co.uk/news/magazine-10725163
It would allow people to participate in the economy and pursue productivity or creative pursuits without worrying about starving.
I doubt it's coming, but I don't see its downside in a world where per capita productivity is trivial.
Not working, or working on figuring out how to get paid in the new environment, when what worked last year is now automated...
Agreed.
Vaguely relevant story time for 1st world crowd: In my country (South Africa) it is illegal for the public to refuel their own cars. There is dedicated staff that does nothing else all day - essentially behave like restaurant waiters but for petrol stations.
Realistically the law is only in place for job creation purposes - it is by most measures unnecessary. And yet I've come to appreciate it: I don't need to leave my car. The people handling flammable liquids do this all day long & received training - much better than random McDonalds munching idiots. Credit card speedpoint gets brought to my car window, windscreen washed, tyre pressure checked etc. I end up (voluntarily) tipping 0.5 USD - a drop in the ocean. I have a million problems so a guy making my life easier there & smiling whilst doing so...I'm totally ok with carrying him on my tax/fuel bill.
You miss my point entirely. It is not useless work. Sure its not economically efficient, but on balance I'd rather have people streamline the refueling process than sit in front of a TV and collect a paycheque.
http://en.wikipedia.org/wiki/Parable_of_the_broken_window
And no it is not one. There is a big difference between "economically efficient" and "wasting their times on a unnecessary job", especially in the context of GMI.
You make it look that these two are the only alternatives when they are not (e.g. people could also be studying or actively searching for a job). Pretty much the definition of false dichotomy isn't it?
Yes and no. You underestimate the experience imo. e.g. Today I had two people dressed in red shirts (for valentines day) focusing on me, my car & what I wanted. Yes you can automate the physical refueling, but can you automate the conversation about the Tshirts specifically bought for Valetines day?
For a minor cost I can:
> Create a job > Interact with a person in a different class > Ease my daily burden > Improve safety > Assist the local economy (micro scale) > Improve social balances
As I said...I tip 0.5USD at petrol stations and I could totally get away with zero.
Yes the mechanical act could be automated but that is entirely besides the point.
Thank you. That is certainly news to me.
I didn't know that part - presumably not lower enough to cover the wages, though, or we'd see it done voluntarily elsewhere.
And to a lesser extent the modern day "professional" military.
Massive jobs program hammers, begging for nails.
https://news.ycombinator.com/item?id=7240235
>Massive jobs program hammers, begging for nails.
Perhaps. I think you judge them unjustly though - it is certainly prone to bureaucrats sinking the scheme with BS, but I don't think it is in itself prone to sinking.
As for the TSA comment - my condolences to all Americans. I don't feel its a representative example of the current debate though.
As I said this is something that makes me deeply uneasy. I'm pretty relaxed about most "serious" issues...e.g. Climate change - I feel "it'll be fine & we'll make it work somehow". Looking at this problem I can't help but feel it will not be OK. And thats coming from a guy in South Africa - a society that is barely hanging on under the strain of inequality. SA is on the very edge of what the social fabric can withstand in terms of inequality & AI will cause inequality that is much worse.
<not serious>
Plainly the best thing to do is hurry the development of machine learning algorithms for active environments, also known as AI, in order to minimize the gap of time between most of the world's people becoming unemployable and most of the world's people being made of atoms the AI can use for something else.
</not serious>
And my apologies for just speaking the truths that many here already know; over my decades in computer science, I have tried to hold my tongue whenever the AI cultists scream at passers-by that the singularity is near -- but every now and again I must scream back for my own sanity.
This article focuses on the positive effects. In their book they actually spend quite a lot of time discussing potential negative effects as well, such as unemployment. You'll be glad to hear that they agree with you that jobs like programming and engineering will be ever more important in that world.
Because the exponential, digital, and recombinant powers of the second machine age have made it possible for humanity to create two of the most important one-time events in our history: the emergence of real, useful artificial intelligence (AI) and the connection of most of the people on the planet via a common digital network.
"The emergence of real, useful artificial intelligence" is what I dispute; if the article chose to describe this as the "emergence of real, useful, algorithms and systems to address meaningful problems" then, fine. But of course, that isn't news, either -- it's merely the story of humanity in a sentence: it is us using our bipedalism and our oversized brains to do what we have always done to survive and thrive.
Once you understand the algorithm, suddenly it doesn't seem mysterious and then it can't be "true" AI. I'll agree that advancements in machine learning aren't necessarily leading to strong AI, but they are still incredible advancements. Many problems in the field of AI have been solved, and then suddenly dismissed as not being "AI".
However I quite agree that the cheerleaders for AI frequently do us all a disservice by exaggerating the successes, and above all by exaggerating how close we may be to a breakthrough.
Humanity's odds aren't good.
It definitely the dominate factor in the future.
Want to compare notes, then? What've you got? Email me.