8 ms·

What does Hacker News think about AI? Is it real this time, or are we in for another winter? I'm seeing a lot of grand claims, and it certainly seems like there

by msvan 10y ago

What does Hacker News think about AI? Is it real this time, or are we in for another winter? I'm seeing a lot of grand claims, and it certainly seems like there are plenty of applications, but I'm still not totally convinced that it will turn the entire economy upside down. Given the enormous amount of press, tweets, blog posts, conferences, degree programs, seminars and interviews popping up it seems like there has to be something more than just hot air here. Still, the most outrageous predictions are hinged on breakthroughs in unsupervised learning happening. Taking the pessimistic view on science, what if we don't get there?

Swizec 10y ago
Unlike 30, 20, or even 10 years ago, we have heaps upon heaps of labeled datasets now. So much human life happens through technology that we are doing an astounding job of producing labeled datasets. And computers are fast now. And RAM is infinite. And GPUs are fast and plentiful. And all of this is cheap. AI stuff is already real more than most of us realize on a day-to-day basis. While machines might not be "intelligent" per se, their cleverness has definitely started impacting desk jobs. The desk job stuff is what most of those alarmist news items are worried about. Nobody seems to worry [too much] about reducing blue collar jobs through automation. Can you imagine how many people you'd need to unload a modern container ship without computers tracking stuff and optimizing storage? Or how much work it would take to harvest modern crops where automated harvesters are used?
- pakl 10y ago
  There are some promising ideas out there based on unsupervised ("self-supervised") learning. There, the problem of needing big labeled datasets doesn't exist: just turn on a camera and have a motor point it at the surrounding world. Data of this kind is and always has been super abundant. But using it requires making a conceptual leap away from supervised pattern recognition -- which is pretty much what today's "AI" does...
  - andirk 10y ago
    It'll make you miss spam.
imh 10y ago
Like they say in the article, there's general AI (think sci-fi computers with minds) and specialized AI (think good old fashioned statistical models, but applied to more things and super effective). Specialized AI (and I hate calling it AI) is coming along really quickly. We're getting better at it in existing fields and learning to apply it to new fields. More than anything, we just have so much data on everything now, and computers are pretty powerful now, so even old school models are finding tons of new applications. Generalized AI is a different story. We are a few really major breakthroughs away. We aren't even 100% sure if they are possible, much less understanding how to do them. These aren't the normal slowly-chip-away-at-it breakthroughs, these are things we have no clue about. With something like that, who can really say how far we are? It could be 5 years, it could be never.
- darawk 10y ago
  Honest question: Why do you think that generalized and specialized AI are distinct things? Is it not possible that general AI is a specialized AI applied over the field of specialized AI generation?
  - daveguy 10y ago
    My two cents: they are separate because there is no current algorithm that can take us from modeling (whether classical statistics or neural net) to intelligence. Applying our current specialized techniques to AI generation has not gotten us there. That is because the techniques are mostly model tweaking techniques. The models are generated and trained for each problem domain. A combined solution may be developed soon, but I doubt it. There was a great article recently on HN that highlights the current problems: http://www.theverge.com/2016/10/10/13224930/ai-deep-learning-limitations-drawbacks http://www.theverge.com/2016/10/10/13224930/ai-deep-learning... https://news.ycombinator.com/item?id=12684417 https://news.ycombinator.com/item?id=12684417 Just because we may acquire the processing power estimated to be used in the brain (in operations per second) doesn't mean we know how to write the software to accomplish the task. It is very clear current algorithms won't cut it. Also, I think we are a few orders of magnitude off on raw processing requirements because I think it is a bandwidth issue as much as an operations per second issue. TL;DR - you could throw as much processing power and data as you want at any current deep NN or their derivatives and you wouldn't get general intelligence. That said I don't think the winter will be as bad as before because, like OP says, specialized AI is useful.
joe_the_user 10y ago
I think machine learning has a real ability to do certain things (image recognition especially) and do them in a way that isn't hype by itself. I suspect that you'll encounter limits to the techniques before they do everything a person can do. If there's a potential problem with machine learning that could sink the enthusiasm over time, I suspect it would come because machine learning applications are these black-boxes system which are the product of training with huge datasets and which use the very tuned-level of expertise of their creators (a common joke is talking of the "graduate school descent" algorithm, getting enough grad students to tune your app till it work). It may be that the deployers of these applications may find that when they have to train them again, in a year's time, that the geniuses have moved on to other things or that the geniuses now charge rates that look excessive for an application that works for just a year. But that's just spinning possibilities. Currently things seem to be going great.
agentgt 10y ago
I have often wandered if there is some sort universe limitation to general AI singularities particularly if the current universe is actually a simulation. An intellectual singularity or extremely rampant aggressive intelligence might be an errant state that gets dealt with (aka universe circuit breaker / watchdog). This might also explain the Fermi Paradox. The other idea I have doted on is that perhaps universes are the manufacturing tool to create super AI by some parent universe. > Taking the pessimistic view on science, what if we don't get there? I have no interest in living forever but I really really wish I could be told what will happen or what did happen or what all is. I'm sure specialized intelligence will continue to improve but my gut says general intelligence is probably not in our lifetime (or is limited or capped from above mentioned pop-culture-probably-wrong reasons).
- imtringued 10y ago
  >The other idea I have doted on is that perhaps universes are the manufacturing tool to create super AI by some parent universe. And perhaps our creators are long dead but the simulation keeps going...
- erikpukinskis 10y ago
  I'm curious: how interested are you in understanding the Earth ecology? Because we understand very little about it, and we are very rapidly destroying the information it contains. I ask because I wonder, if we did meet or make a creature which could learn everything, if it wouldn't say some variation on "99.9999% of everything there is to know is right there in your ecology and visible to the naked eye. Go look." I also wonder if one of the first things the sentient AIs teach us is that yes, we are committing an egregious ongoing information Holocaust through habitat destruction.
pakl 10y ago
I'm optimistic about what'll happen right after the upcoming AI winter. :) So long as AI remains tasked with categorizing human-taken photos or playing human-created games (no matter how "complex"), AI will remain just that -- artificial -- and not "real". Given the rate of hype that you point out, I suspect a winter of some kind will hit before enough folks realize this. [edited: reorder]
- ktRolster 10y ago
  What should they focus on then?
  - pakl 10y ago
    If one wants general AI that can deal with/understand the world, the system needs to learn based on (raw, unadulterated) data from the world. These data are highly dynamic and rarely fit neatly into human-labeled categories. This is part of why the currently-hyped supervised pattern recognition is not all that helpful for general AI. By the way, the above requirement of real world data probably even applies to building chatbots. (A successful chatbot will need some understanding of the world it is talking about; this is what researchers mean when they say "grounding" is important for NLP).
AJRF 10y ago
"press, tweets, blog posts, conferences, degree programs, seminars and interviews" That is the hot air
asimuvPR 10y ago
There is a real risk related to goal oriented AI. It does not need to feel or dream. Merely having survival as its goal is sufficient to make it dangerous to other life forms. Worse is that it can happen at any time (it may have happened already). Given the computing power, tools, and availability of knowledge we can assume that it can be done outside of a controlled lab environment by a non-scientist.
- andirk 10y ago
  Huge point you make that it can be done by non-scientists now. Now we can all play with fire. That much power in idle hands controlled by mediocre minds is like everyone being magical.
  - mr_spothawk 10y ago
    "mediocre minds"?
    - asimuvPR 10y ago
      It sounds weird, but I believe its meant as "untrained minds".
      - andirk 10y ago
        Yes untrained but more broad to mean people who don't understand the magic, people who are negligent, etc. I kind of got it from Einstein's quote, "Great spirits have always encountered opposition from mediocre minds."
- erikpukinskis 10y ago
  I'm not too worried about it, because AIs will need to feel and dream and care and suffer and empathize to begin to be as intelligent as a human. Imagine a human who lacks empathy entirely. That's a disability. They may be able to do some amount of destruction, but at some scale they simply lack the social intelligence necessary to compete with the entire species. This is the most common mistake I think people make when reasoning about AI: they think human limitations are weaknesses. But they're not weaknesses they're tradeoffs. Natural selection has had a chance to reward all kinds of variations, including more cortex, and less empathy. But we ended up where we are because of tradeoffs. Any AI which is intelligent in the same way humans are will also have our limitations. Any AI which doesn't have our limitations won't be as smart as us in those respects. You have to really ask yourself what the difference is between a human with an AI simulator and an AI with a human simulator. In the limit of simulator quality there is none.
l33tbro 10y ago
As an Ai "hipster" (been into it before it was cool), it's certainly interesting seeing the mainstream culture now catching up to the implications it has. I'm glad this is happening, and I hope the discussion broadens and we hear from all people. I'm one of the whacky ones who believes it's a really important rung in our evolution as a species, so I think being collectively aware of where we're going can only be a good thing.

by msvan 10y ago

Swizec 10y ago

Unlike 30, 20, or even 10 years ago, we have heaps upon heaps of labeled datasets now. So much human life happens through technology that we are doing an astounding job of producing labeled datasets. And computers are fast now. And RAM is infinite. And GPUs are fast and plentiful. And all of this is cheap. AI stuff is already real more than most of us realize on a day-to-day basis. While machines might not be "intelligent" per se, their cleverness has definitely started impacting desk jobs. The desk job stuff is what most of those alarmist news items are worried about. Nobody seems to worry [too much] about reducing blue collar jobs through automation. Can you imagine how many people you'd need to unload a modern container ship without computers tracking stuff and optimizing storage? Or how much work it would take to harvest modern crops where automated harvesters are used?

pakl 10y ago
There are some promising ideas out there based on unsupervised ("self-supervised") learning. There, the problem of needing big labeled datasets doesn't exist: just turn on a camera and have a motor point it at the surrounding world. Data of this kind is and always has been super abundant. But using it requires making a conceptual leap away from supervised pattern recognition -- which is pretty much what today's "AI" does...
- andirk 10y ago
  It'll make you miss spam.

imh 10y ago

Like they say in the article, there's general AI (think sci-fi computers with minds) and specialized AI (think good old fashioned statistical models, but applied to more things and super effective). Specialized AI (and I hate calling it AI) is coming along really quickly. We're getting better at it in existing fields and learning to apply it to new fields. More than anything, we just have so much data on everything now, and computers are pretty powerful now, so even old school models are finding tons of new applications. Generalized AI is a different story. We are a few really major breakthroughs away. We aren't even 100% sure if they are possible, much less understanding how to do them. These aren't the normal slowly-chip-away-at-it breakthroughs, these are things we have no clue about. With something like that, who can really say how far we are? It could be 5 years, it could be never.

darawk 10y ago
Honest question: Why do you think that generalized and specialized AI are distinct things? Is it not possible that general AI is a specialized AI applied over the field of specialized AI generation?
- daveguy 10y ago
  My two cents: they are separate because there is no current algorithm that can take us from modeling (whether classical statistics or neural net) to intelligence. Applying our current specialized techniques to AI generation has not gotten us there. That is because the techniques are mostly model tweaking techniques. The models are generated and trained for each problem domain. A combined solution may be developed soon, but I doubt it. There was a great article recently on HN that highlights the current problems: http://www.theverge.com/2016/10/10/13224930/ai-deep-learning-limitations-drawbacks http://www.theverge.com/2016/10/10/13224930/ai-deep-learning... https://news.ycombinator.com/item?id=12684417 https://news.ycombinator.com/item?id=12684417 Just because we may acquire the processing power estimated to be used in the brain (in operations per second) doesn't mean we know how to write the software to accomplish the task. It is very clear current algorithms won't cut it. Also, I think we are a few orders of magnitude off on raw processing requirements because I think it is a bandwidth issue as much as an operations per second issue. TL;DR - you could throw as much processing power and data as you want at any current deep NN or their derivatives and you wouldn't get general intelligence. That said I don't think the winter will be as bad as before because, like OP says, specialized AI is useful.

joe_the_user 10y ago

I think machine learning has a real ability to do certain things (image recognition especially) and do them in a way that isn't hype by itself. I suspect that you'll encounter limits to the techniques before they do everything a person can do. If there's a potential problem with machine learning that could sink the enthusiasm over time, I suspect it would come because machine learning applications are these black-boxes system which are the product of training with huge datasets and which use the very tuned-level of expertise of their creators (a common joke is talking of the "graduate school descent" algorithm, getting enough grad students to tune your app till it work). It may be that the deployers of these applications may find that when they have to train them again, in a year's time, that the geniuses have moved on to other things or that the geniuses now charge rates that look excessive for an application that works for just a year. But that's just spinning possibilities. Currently things seem to be going great.

agentgt 10y ago

I have often wandered if there is some sort universe limitation to general AI singularities particularly if the current universe is actually a simulation. An intellectual singularity or extremely rampant aggressive intelligence might be an errant state that gets dealt with (aka universe circuit breaker / watchdog). This might also explain the Fermi Paradox. The other idea I have doted on is that perhaps universes are the manufacturing tool to create super AI by some parent universe. > Taking the pessimistic view on science, what if we don't get there? I have no interest in living forever but I really really wish I could be told what will happen or what did happen or what all is. I'm sure specialized intelligence will continue to improve but my gut says general intelligence is probably not in our lifetime (or is limited or capped from above mentioned pop-culture-probably-wrong reasons).

imtringued 10y ago
>The other idea I have doted on is that perhaps universes are the manufacturing tool to create super AI by some parent universe. And perhaps our creators are long dead but the simulation keeps going...
erikpukinskis 10y ago
I'm curious: how interested are you in understanding the Earth ecology? Because we understand very little about it, and we are very rapidly destroying the information it contains. I ask because I wonder, if we did meet or make a creature which could learn everything, if it wouldn't say some variation on "99.9999% of everything there is to know is right there in your ecology and visible to the naked eye. Go look." I also wonder if one of the first things the sentient AIs teach us is that yes, we are committing an egregious ongoing information Holocaust through habitat destruction.

pakl 10y ago

I'm optimistic about what'll happen right after the upcoming AI winter. :) So long as AI remains tasked with categorizing human-taken photos or playing human-created games (no matter how "complex"), AI will remain just that -- artificial -- and not "real". Given the rate of hype that you point out, I suspect a winter of some kind will hit before enough folks realize this. [edited: reorder]

ktRolster 10y ago
What should they focus on then?
- pakl 10y ago
  If one wants general AI that can deal with/understand the world, the system needs to learn based on (raw, unadulterated) data from the world. These data are highly dynamic and rarely fit neatly into human-labeled categories. This is part of why the currently-hyped supervised pattern recognition is not all that helpful for general AI. By the way, the above requirement of real world data probably even applies to building chatbots. (A successful chatbot will need some understanding of the world it is talking about; this is what researchers mean when they say "grounding" is important for NLP).

AJRF 10y ago

"press, tweets, blog posts, conferences, degree programs, seminars and interviews" That is the hot air

asimuvPR 10y ago

There is a real risk related to goal oriented AI. It does not need to feel or dream. Merely having survival as its goal is sufficient to make it dangerous to other life forms. Worse is that it can happen at any time (it may have happened already). Given the computing power, tools, and availability of knowledge we can assume that it can be done outside of a controlled lab environment by a non-scientist.

andirk 10y ago
Huge point you make that it can be done by non-scientists now. Now we can all play with fire. That much power in idle hands controlled by mediocre minds is like everyone being magical.
- mr_spothawk 10y ago
  "mediocre minds"?
  - asimuvPR 10y ago
    It sounds weird, but I believe its meant as "untrained minds".
    - andirk 10y ago
      Yes untrained but more broad to mean people who don't understand the magic, people who are negligent, etc. I kind of got it from Einstein's quote, "Great spirits have always encountered opposition from mediocre minds."
erikpukinskis 10y ago
I'm not too worried about it, because AIs will need to feel and dream and care and suffer and empathize to begin to be as intelligent as a human. Imagine a human who lacks empathy entirely. That's a disability. They may be able to do some amount of destruction, but at some scale they simply lack the social intelligence necessary to compete with the entire species. This is the most common mistake I think people make when reasoning about AI: they think human limitations are weaknesses. But they're not weaknesses they're tradeoffs. Natural selection has had a chance to reward all kinds of variations, including more cortex, and less empathy. But we ended up where we are because of tradeoffs. Any AI which is intelligent in the same way humans are will also have our limitations. Any AI which doesn't have our limitations won't be as smart as us in those respects. You have to really ask yourself what the difference is between a human with an AI simulator and an AI with a human simulator. In the limit of simulator quality there is none.

l33tbro 10y ago

As an Ai "hipster" (been into it before it was cool), it's certainly interesting seeing the mainstream culture now catching up to the implications it has. I'm glad this is happening, and I hope the discussion broadens and we hear from all people. I'm one of the whacky ones who believes it's a really important rung in our evolution as a species, so I think being collectively aware of where we're going can only be a good thing.