4 ms·

For a field like AI, it's extremely difficult to decide if knee-jerk aversion to specific advancements is something worth taking seriously or not. For a field w

by gfodor 9y ago

For a field like AI, it's extremely difficult to decide if knee-jerk aversion to specific advancements is something worth taking seriously or not. For a field where the entire point is to replace humans in certain functions, it's inevitable that there will be leaps forward that people are extremely uncomfortable with in the short term. This may be an example of something where things need to be altered, or it may not. But in either case it seems short-sighted to just think this question can be resolved in 24 hours after a demo due to a few hastily-written tweets and thrown-together articles. What's important is that as AI continues to encroach upon human capabilities that there be a framework for navigating these transitions and knowing where the lines are. Hopefully as this kind of thing becomes more common public opinion will shift from caring about the creepy-demo-of-the-week to a wider scope of how we grapple as a society with these kinds of advancements. Emotional reactions and ad-hoc decisionmaking based on mob rule is going to result in a random, arbitrary outcome which will almost certainly be suboptimal.

codingdave 9y ago
You can both take something seriously, and also deliberately decide that the reactions were not sufficient to invoke a change of direction. Being serious about evaluating a customer response doesn't mean you cater to that response.
- salawat 9y ago
  Again, you can. See how well that worked for NASA's Integrated Program Plan. It was too ambitious a project that generated some spectacular design and planning work, but ended up dashed on the rocks of the political and social environment of the time. Learn from the past that you might not repeat it.
salawat 9y ago
Sometimes suboptimal is a lot less frightening and destructive than optimal. Dictatorships and despotism are the optimal solution for wielding of executive power. Murder is the optimal means through which to immediately resolve problems created by other people from a time investment point of view. Rape is an optimal solution to spreading genetic material. See the pattern? Knee-jerk reactions from the layperson AREN'T necessarily bad. If anything, they should be the first sign that a researcher needs to back up from that awesome thing you just did, and take off the razor thin beam of relevance you used to achieve your breakthrough, and begin the HARD part of research. Turning it from an interesting academic factoid into a societal good. When we do research, we are free to ignore huge swathes of the world to be able to reach or findings. The reckoning, however, comes when we have to ask ourselves 'how prepared is the world for this?' The story of Doctor Frankenstein isn't just some scary story for children. It should be haunting a researcher night and day. What am I enabling? How will society take this? It shouldn't stop you by all means, but it will keep you grounded and aware of the fact your research isn't happening in a void. Remember, all those bits and pieces you are putting out of your mind are the bread and butter of every layperson's day. Being a researcher doesn't absolve you of your status as a member of society. It is YOUR responsibility to take your idea from the theoretical to the practical. The rest of society is busy staying intact enough to support your research efforts.
- gfodor 9y ago
  I'm not sure who you're arguing with here? My point was that a suboptimal solution is one where there was no careful, methodical thought to the design of AI systems. If AI designers and researchers blindly react to public outcry after 24 hours from a demo (which, in general, is something I would expect Google to do) the kind of thinking you mention above is just as unlikely to happen as if they trudged on forward without any consideration to these things at all. In both cases, this is a suboptimal outcome for society. In one, we get a fairly random, undesigned world of AI systems that don't serve anyone well and generally are underutilized because nobody is willing to push the boundaries. In the other, we get dystopian AI hell. It's important that researchers be given the space to think you describe. The right way to allow that is to foster an informed and open-minded public about the emergence of AI and the decisions we need to make about it as a society, and not have prominent voices writing about knee-jerk reactions making authoritative demands to a specific public demo after just a day. (See Max Tegmark's book Life 3.0 for the kinds of stuff we need more of being put into the world imho.)
  - salawat 9y ago
    It is this part I"m referring to: >Emotional reactions and ad-hoc decisionmaking based on mob rule is going to result in a random, arbitrary outcome which will almost certainly be suboptimal. I may have initially misinterpreted some context when I read your comment before, though my conclusions haven't changed much. I agree with you that AI research must be methodically and transparently approached. The part where I was trying to speak toward, however, was that AI researchers/designers are going to necessarily be constrained by how much faith and goodwill a (currently) troubled society, plagued with inequality and hardship in the face of increasing job loss to automation. This is a feature, not a bug to be dismissed. AI is akin to nuclear technology in that it is perceived as a huge game changer. One that brings with it a huge potential for societal change and upheaval. The AI designer/researcher DOESN'T necessarily see it that way. It is just a tool to do a thing. The populace, however, don't grasp the limits the researcher/designer does. They are being asked to give the same degree of faith to the AI field that nuclear physics was given in the last century. Combine that with a severe distrust in the motives of the organizations doing the actual development and you have the stage set for large-scale rejection of AI research. You said it yourself before. AI is a field that is trying to replace humans (having to do stuff). That is not a viewpoint that will win you the favor of the populace. No one will get further than the first half of the sentence before the hackles go up. Nobody trusts a Utopia where people don't have to work because machines do it for them. Nor will they trust the men behind the machines. It's sad, but it is the kind of thinking that has kept humanity OUT of a complete dystopian hellholes thus far. I don't know the fix, but I do know that if we can't solve some very fundamental social problems that advances in artificial intelligence and machine learning are poised to make worse, you will find the field of study becoming more and more difficult to find public acceptance in. The only inkling of a clue I have is it may have to do with the "industrial asymmetry" of our current times. If you don't take heed of the state of society, you may end up in the same place NASA was for the latter half of the 20th Century. BEAUTIFUL theoretical groundwork laid, but hamstrung in execution by popular sentiment. (See some documentaries to learn how "thrilled" many were with the Space Race. NASA still hasn't recovered from the backlash despite some of the best science I've ever seen)

by gfodor 9y ago

codingdave 9y ago

You can both take something seriously, and also deliberately decide that the reactions were not sufficient to invoke a change of direction. Being serious about evaluating a customer response doesn't mean you cater to that response.

salawat 9y ago
Again, you can. See how well that worked for NASA's Integrated Program Plan. It was too ambitious a project that generated some spectacular design and planning work, but ended up dashed on the rocks of the political and social environment of the time. Learn from the past that you might not repeat it.

salawat 9y ago

Sometimes suboptimal is a lot less frightening and destructive than optimal. Dictatorships and despotism are the optimal solution for wielding of executive power. Murder is the optimal means through which to immediately resolve problems created by other people from a time investment point of view. Rape is an optimal solution to spreading genetic material. See the pattern? Knee-jerk reactions from the layperson AREN'T necessarily bad. If anything, they should be the first sign that a researcher needs to back up from that awesome thing you just did, and take off the razor thin beam of relevance you used to achieve your breakthrough, and begin the HARD part of research. Turning it from an interesting academic factoid into a societal good. When we do research, we are free to ignore huge swathes of the world to be able to reach or findings. The reckoning, however, comes when we have to ask ourselves 'how prepared is the world for this?' The story of Doctor Frankenstein isn't just some scary story for children. It should be haunting a researcher night and day. What am I enabling? How will society take this? It shouldn't stop you by all means, but it will keep you grounded and aware of the fact your research isn't happening in a void. Remember, all those bits and pieces you are putting out of your mind are the bread and butter of every layperson's day. Being a researcher doesn't absolve you of your status as a member of society. It is YOUR responsibility to take your idea from the theoretical to the practical. The rest of society is busy staying intact enough to support your research efforts.

gfodor 9y ago
I'm not sure who you're arguing with here? My point was that a suboptimal solution is one where there was no careful, methodical thought to the design of AI systems. If AI designers and researchers blindly react to public outcry after 24 hours from a demo (which, in general, is something I would expect Google to do) the kind of thinking you mention above is just as unlikely to happen as if they trudged on forward without any consideration to these things at all. In both cases, this is a suboptimal outcome for society. In one, we get a fairly random, undesigned world of AI systems that don't serve anyone well and generally are underutilized because nobody is willing to push the boundaries. In the other, we get dystopian AI hell. It's important that researchers be given the space to think you describe. The right way to allow that is to foster an informed and open-minded public about the emergence of AI and the decisions we need to make about it as a society, and not have prominent voices writing about knee-jerk reactions making authoritative demands to a specific public demo after just a day. (See Max Tegmark's book Life 3.0 for the kinds of stuff we need more of being put into the world imho.)
- salawat 9y ago
  It is this part I"m referring to: >Emotional reactions and ad-hoc decisionmaking based on mob rule is going to result in a random, arbitrary outcome which will almost certainly be suboptimal. I may have initially misinterpreted some context when I read your comment before, though my conclusions haven't changed much. I agree with you that AI research must be methodically and transparently approached. The part where I was trying to speak toward, however, was that AI researchers/designers are going to necessarily be constrained by how much faith and goodwill a (currently) troubled society, plagued with inequality and hardship in the face of increasing job loss to automation. This is a feature, not a bug to be dismissed. AI is akin to nuclear technology in that it is perceived as a huge game changer. One that brings with it a huge potential for societal change and upheaval. The AI designer/researcher DOESN'T necessarily see it that way. It is just a tool to do a thing. The populace, however, don't grasp the limits the researcher/designer does. They are being asked to give the same degree of faith to the AI field that nuclear physics was given in the last century. Combine that with a severe distrust in the motives of the organizations doing the actual development and you have the stage set for large-scale rejection of AI research. You said it yourself before. AI is a field that is trying to replace humans (having to do stuff). That is not a viewpoint that will win you the favor of the populace. No one will get further than the first half of the sentence before the hackles go up. Nobody trusts a Utopia where people don't have to work because machines do it for them. Nor will they trust the men behind the machines. It's sad, but it is the kind of thinking that has kept humanity OUT of a complete dystopian hellholes thus far. I don't know the fix, but I do know that if we can't solve some very fundamental social problems that advances in artificial intelligence and machine learning are poised to make worse, you will find the field of study becoming more and more difficult to find public acceptance in. The only inkling of a clue I have is it may have to do with the "industrial asymmetry" of our current times. If you don't take heed of the state of society, you may end up in the same place NASA was for the latter half of the 20th Century. BEAUTIFUL theoretical groundwork laid, but hamstrung in execution by popular sentiment. (See some documentaries to learn how "thrilled" many were with the Space Race. NASA still hasn't recovered from the backlash despite some of the best science I've ever seen)