I love SciFi. I wanted to be Spock; I wanted to be The Doctor. I wanted to live in Caves of Steel and be the one person friendly with robots. For that matter, sometimes I wanted to be Susan Calvin. When I was a child. As a consequence, since I was a child (1982, I was 17) I've studied formally psychology and AI (getting two degrees in each) and spent seven years practicing programming and systems design in industry. And since becoming faculty in 2002 I've directed most of my research (and personal) time to understanding the evolution of cognition and culture. And since 2008 that work has extended to studying the creation and destruction of public goods – because knowledge in culture, the basis of humans' exceptional intelligence, is a public good. In the last few years I've been fortunate to be asked to apply my knowledge to public policy, and in so doing I've been learning a lot about law, ethics, and legislation too.
Rights are a protective force of last resort. Rights are one of the ways you try to defend other people like yourself, and you do that because one day you or someone you love may only be able to defend themselves that way. They're a means of persuasion, of organising group support. When you stop identifying with other people, you start denying them rights. In the limit case, you do things like stripping and raping them after having throwing their baby into the burning remnants of their home. Or calling airstrikes against your own citizens if they don't vote for your political party. People are doing these things to each other right now. We can be sufficiently polarised that we spend decades not speaking to our neighbours, not following the same religion or learning the same language, and then when push comes to shove (say we are stressed by increasingly severe and frequent drought or hurricanes) we behave as if other people just like ourselves have no rights at all.
Some people are worrying about whether we need to allocate rights to robots. So how is a robot a thing like yourself you might need to defend? Well, if you think what makes you important or special is having a lot of knowledge, or doing something super creative, then you might think a robot could potentially be as much like you as at least some other people are. In fact, you might think that robots need more protection than you do which would be nice because you know you can't really protect yourself. You are going to die in less than 100 years, but maybe you could make a robot that would live forever, so it would deserve more rights and resources than you (or any other human) ever could.
If you really cared about the well being of AI, you would want AI systems to be easy to understand and maintain. You wouldn't try to encourage companies to make their systems obfuscated by telling them they have less liability if their machine intelligence is opaque, or worse, 'conscious'. If you want to protect the knowledge and creativity we build into our artefacts, then work to make the corporations and institutions that manufacture and use those artefacts more stable and resilient. Work on how we can make AI more maintainable, and how we can make the individuals and corporations that use AI to alter our lives more accountable. Work on how we can make society sustainable. Work on how we can reduce political polarisation, so that we don't just wave beliefs around like flags but actually use them for real thinking.
I feel I'm writing this post over and over, but maybe if I say it more bluntly it will make more sense. Stop playing Dada with AI. Creativity is a wonderful thing, but now is not the time to be confusing democratic publics about the nature of communication technology. AI is helping people dismantle the institutions that keep us safe, like public health care, the US department of energy, the EU / NATO / OECD. It is far too often being used to hand more power to the very few, because that is easier than making the whole world safer.
If you want your ideas to be safe, make sure you back up your phones, laptops, and robots. And preferably, write your ideas down on something analog, maybe on a variety of different materials. But nothing will last forever. Don't promise people a new immortality as ungrounded in the laws of physics, biology, and computation as the old form of immortality you gave up on when you were 14.
 |
| Human rights are precious but fragile. |
Some people are worrying about whether we need to allocate rights to robots. So how is a robot a thing like yourself you might need to defend? Well, if you think what makes you important or special is having a lot of knowledge, or doing something super creative, then you might think a robot could potentially be as much like you as at least some other people are. In fact, you might think that robots need more protection than you do which would be nice because you know you can't really protect yourself. You are going to die in less than 100 years, but maybe you could make a robot that would live forever, so it would deserve more rights and resources than you (or any other human) ever could.
AI does not need to need rights
I frankly think this is sick. AI is not a form of math; it's a form of computation. Computation is a physical process that takes time, energy, and space. It and its outputs do not last forever. You get books that last a few thousand years in exceptional circumstances, you may get stories that last far longer than that. But no mechanical artefact is at all likely to last as long as a person is likely to last. Even special artefacts like languages, or nations, only last recognisably for a few hundred years.If you really cared about the well being of AI, you would want AI systems to be easy to understand and maintain. You wouldn't try to encourage companies to make their systems obfuscated by telling them they have less liability if their machine intelligence is opaque, or worse, 'conscious'. If you want to protect the knowledge and creativity we build into our artefacts, then work to make the corporations and institutions that manufacture and use those artefacts more stable and resilient. Work on how we can make AI more maintainable, and how we can make the individuals and corporations that use AI to alter our lives more accountable. Work on how we can make society sustainable. Work on how we can reduce political polarisation, so that we don't just wave beliefs around like flags but actually use them for real thinking.
I feel I'm writing this post over and over, but maybe if I say it more bluntly it will make more sense. Stop playing Dada with AI. Creativity is a wonderful thing, but now is not the time to be confusing democratic publics about the nature of communication technology. AI is helping people dismantle the institutions that keep us safe, like public health care, the US department of energy, the EU / NATO / OECD. It is far too often being used to hand more power to the very few, because that is easier than making the whole world safer.
If you want your ideas to be safe, make sure you back up your phones, laptops, and robots. And preferably, write your ideas down on something analog, maybe on a variety of different materials. But nothing will last forever. Don't promise people a new immortality as ungrounded in the laws of physics, biology, and computation as the old form of immortality you gave up on when you were 14.
Humans do not need to need AI to have rights
The one argument not covered above is that AI perhaps does not need rights in itself, but because we constantly confuse AI with humanity, we need to treat it as we would like humanity to be treated. This argument comes from Kant, and is most recently made by Gunkel.
We addressed this argument in 2010. Of course Kant is right. But if you don't want people to be hurt by AI being treated as not human, make it obvious to them that AI really is not human. Again, if you really cared, you could address this. In fact, even if you worked in media studies rather than AI or psychology or politics, you could address this. Because the kinds of things you can actually learn from media studies is what people think when and how to manipulate that and them. So how about doing that in a way that addresses real, concrete problems with justice, economics, and democracy?
Slightly less blunt versions of the above argument
- Why (or rather, when) suffering in AI is incoherent.
- If robots ever need rights we'll have designed them unjustly
- Robots are more like novels than children
- What Makes a Person? Five Reasons Not to Other AI
An equally blunt academic version of the above argument aimed at the European Parliament and Commission
Personally I think the Lacuna article as at least as important as the Science article about machine and human bias. But law and ethics are by their nature less transparently evident than science. I only hope our article will make the impact it needs to, because otherwise we all will lose the benefits of tax and legal liability against the corporations that make AI and robots, and gain nothing but badly-designed robots, and more fantasies through which we can be exploited.
I shouldn't say "I only hope." I also put a lot of time into trying to make sure people read and understand it. I hope you do. Please use the comment section here if you have any questions.
Addendum (24 October) this post made some people question whether I believed humans have rights. Sorry, I took it as obvious that human rights are what I was trying to defend, but nothing is really obvious. Please see the next blogpost which is about human rights.
I shouldn't say "I only hope." I also put a lot of time into trying to make sure people read and understand it. I hope you do. Please use the comment section here if you have any questions.
Addendum (24 October) this post made some people question whether I believed humans have rights. Sorry, I took it as obvious that human rights are what I was trying to defend, but nothing is really obvious. Please see the next blogpost which is about human rights.
Comments