3 ms·

In other words, we'd use an always-connected device (like a phone or a watch) to help us comply with micro-directives. I can totally see some folks rebelling ag

by hosh 11y ago

In other words, we'd use an always-connected device (like a phone or a watch) to help us comply with micro-directives. I can totally see some folks rebelling against that. There's already fear related to strong AIs. On the other hand, this is _exactly_ the same class of problems Google and other self-driving car makers face when it comes to questions of ethics and morality. There were several recent articles on HN about that. The question posed was: should your self-driving car swerve to avoid those kids who just got off the school bus, sacrificing you and the passengers? That leads to an implicit question: should Google and other engineers be responsible for writing such algorithms? What these two law professors are proposing gives a different option. Change the way policy is made. Such micro-directives (which requires a computing device to execute) would include the behavior of self-driving cars as a general case, rather than a special case. The paper might talk about getting the benefits of rules and standards without the cost, but we're really talking about making our legislation legible for machines to follow. On an even bigger scale: how do micro-directives change how we conceive of and think of a democracy or a republic? Are policy makers sufficiently versed in engineering to reason through the impact of the laws? But then again, the idea was to use Big Data methods to collect that. Does that require specialists to design and deploy that system? Wasn't the original idea behind the American form of democracy, a literate voting public that can reason and discuss the impact of policy and make informed votes for those that represent them? Are Big Data methods more accurate representation of the will of the people? Or less? Now, I've only read the first few pages, but I kinda wonder know if the paper will mention any this (and in particular the self-driving car example). Gonna have something to talk with my law geek buddies for a while.

spdionis 11y ago
> a literate voting public that can reason and discuss the impact of policy and make informed votes for those that represent them? I think we already lost that long ago. For a lot of important topics having a literate general public is NP-Hard. That's the main issue with democracy in our age.
- hosh 11y ago
  I agree that we lost that a long time ago. I think we still cling to that ideal as wishful thinking, and that gap between what we'd like democracy to be and what it is now introduces distortions in how we talk about this or reason about it. I'm not entirely sure the micro-directives will help with this. It's something I might meditate upon some time down the road.
digi_owl 11y ago
> The question posed was: should your self-driving car swerve to avoid those kids who just got off the school bus, sacrificing you and the passengers? That strikes me as seeing the car computer as having the same limitations as a human driver. More likely the car computer has long since recognized the bus as being, well, a bus, and thus slowed down to provide shorter breaking distance etc. These are the same exact rules that will be drilled into any driver if they get a sane course before being issued the license. Keep in mind that right now what is regulating speed limits etc, is human reaction times. Once held down, breaks can make a car stop very quickly. But the actual breaking distance is just a fraction of the distance needed for a human driven car to stop. Most of that distances comes from how long it takes for the human brain to go through the steps of noticing, recognizing, and acting on something. Likely any computer driven car out there will be the most defensive driver on the road. Frankly more vehicle related deaths comes from driver hubris than anything else.
- hosh 11y ago
  What you say has merit, yet it sidesteps and ignores the deeper, more troubling issues: is it possible to go Asimov Three Laws of Robotics? Is it possible to specify ethical and moral coding precisely, programmatically? That car example is just a stand-in for an entire class of ethical and moral dilemmas coming out of it. Let's look at it from a different angle. There is a hospital in Oregon who are trying to squeeze more profits, so much so that they induced doctors working there to unionize. The hospital promptly dropped their attempt at outsourcing. It won't be last attempt at this. As the pressure for profits increases, and as machine learning technology gets more advanced, there will come a point when it is the AI that makes the judgement call on who lives or dies. Who is going to program that? Are software engineers necessarily all that better in creating systems that computes the optimal outcome that includes both ethics and morals, than say, lawmakers or philosophers? Are you really OK with having moral choices taken out of your hand and given to a computer, assuming that whomever wrote the code actually knew what they were doing and didn't make mistakes? From this frame, what you are suggesting is that a properly designed computer is better at making moral and ethical judgements than humans.
  - 11y ago

by hosh 11y ago

spdionis 11y ago

> a literate voting public that can reason and discuss the impact of policy and make informed votes for those that represent them? I think we already lost that long ago. For a lot of important topics having a literate general public is NP-Hard. That's the main issue with democracy in our age.

hosh 11y ago
I agree that we lost that a long time ago. I think we still cling to that ideal as wishful thinking, and that gap between what we'd like democracy to be and what it is now introduces distortions in how we talk about this or reason about it. I'm not entirely sure the micro-directives will help with this. It's something I might meditate upon some time down the road.

digi_owl 11y ago

> The question posed was: should your self-driving car swerve to avoid those kids who just got off the school bus, sacrificing you and the passengers? That strikes me as seeing the car computer as having the same limitations as a human driver. More likely the car computer has long since recognized the bus as being, well, a bus, and thus slowed down to provide shorter breaking distance etc. These are the same exact rules that will be drilled into any driver if they get a sane course before being issued the license. Keep in mind that right now what is regulating speed limits etc, is human reaction times. Once held down, breaks can make a car stop very quickly. But the actual breaking distance is just a fraction of the distance needed for a human driven car to stop. Most of that distances comes from how long it takes for the human brain to go through the steps of noticing, recognizing, and acting on something. Likely any computer driven car out there will be the most defensive driver on the road. Frankly more vehicle related deaths comes from driver hubris than anything else.

hosh 11y ago
What you say has merit, yet it sidesteps and ignores the deeper, more troubling issues: is it possible to go Asimov Three Laws of Robotics? Is it possible to specify ethical and moral coding precisely, programmatically? That car example is just a stand-in for an entire class of ethical and moral dilemmas coming out of it. Let's look at it from a different angle. There is a hospital in Oregon who are trying to squeeze more profits, so much so that they induced doctors working there to unionize. The hospital promptly dropped their attempt at outsourcing. It won't be last attempt at this. As the pressure for profits increases, and as machine learning technology gets more advanced, there will come a point when it is the AI that makes the judgement call on who lives or dies. Who is going to program that? Are software engineers necessarily all that better in creating systems that computes the optimal outcome that includes both ethics and morals, than say, lawmakers or philosophers? Are you really OK with having moral choices taken out of your hand and given to a computer, assuming that whomever wrote the code actually knew what they were doing and didn't make mistakes? From this frame, what you are suggesting is that a properly designed computer is better at making moral and ethical judgements than humans.
- 11y ago