While it is popular to rail against the horrors of regulation, copyright laws are rather critical to creators and owners of creations. On the side of good, these laws protect creators and owners from having their works stolen. On the side of evil, these laws can lock creations out of the public domain long after they should have been set free. However, this essay is not aimed at arguing about copyrights as such. Rather, my aim is considering the minor issue of whether Artificial Intelligence (AI) could result in copyright violations. The sort of AI I am considering here is the “classic” sci-fi sort of AI, that is something on par with HAL 9000, C3PO or Data. I am not considering the marketing version of AI, which seems to be just about any sort of thing that does some things. Or does not do them, depending on which cosmic forces are in a pissy mood.
On the face of it, it is rather easy to show that classic AI systems would violate copyright law—at least in some cases. While copyright statements vary, a stock version looks like this:
All rights reserved. No part of this publication may be reproduced, distributed, or transmitted in any form or by any means, including photocopying, recording, or other electronic or mechanical methods, without the prior written permission of the publisher, except in the case of brief quotations embodied in critical reviews and certain other noncommercial uses permitted by copyright law.
The key part is, of course, the bit about reproducing any part of the work by other electronic or mechanical methods. A classic AI system will presumably be electronic (or mechanical, if one wants to go the Difference Engine path) and will probably have a memory system analogous to that of a computer. That is, something like RAM for working memory and something like a drive for long term memory. As such, an AI system would seem to violate copyright law when it reads a copyrighted book or consumes other types of copyrighted media.
One obvious reply to this concern is that a human being is also an electronic system that can reproduce copyrighted works. For example, I can memorize a passage from a book or the lyrics of a song—thus reproducing them in my brain or Cartesian ectoplasm or whatever my mind might be. But, of course, if copyright laws prevented humans from reading books, then there would be little point to it—few would legally buy things that they would be legally forbidden to read. The same would apply to other media,
Obviously enough, copyright law does not forbid humans from consuming such works and a reasonable explanation is that while the human mind can reproduce works, it is generally rather bad at doing so. For example, few people could reproduce even an entire paragraph from a book exactly without considerable practice. As such, one possible reason that copyright laws do not forbid humans from consuming copyrighted media is that the reproduction is imperfect and, for the most part, a human could not reproduce a lengthy work from memory. But, of course, the most obvious reason is that humans generally do not think that when they read a book they are functioning as a reproduction system—they are reproducing the book in their mind.
AI systems of the “classic” sort would differ from humans in many ways, one of which is that they would presumably be capable of perfectly recording copyrighted works, just as a “dumb” computer or smartphone can today. Roughly put, when an AI reads a copyrighted book, it would be analogous to scanning and storing each page of the book—a seemingly clear violation of copyright. The same could be done with copyrighted material in other media, such as music and movies. With such memory, an AI would also be able to reproduce the work exactly—for example, repeating an entire book word for word. To use an analogy, the smart part of the AI would be like a human reading a book and the long-term memory system of the AI would be like a human using a scanner to copy a copyrighted book to a hard drive—a clear copyright violation.
One possibility, which could be yet another reason that AI will kill us all, is that AI systems will be forbidden from viewing copyright works without permission. Alternatively, they could have permission to consume such works and maintain a copy as part of the purchase price. After all, when a human buys a book they get to keep that copy. There would, of course, be a problem with events like a play or a movie in a theater—the AI would, in effect, get to view the movie in the theatre and have a recording of it. This could be offset by including a copy of the movie in the ticket price for everyone, having the AI erase the movie afterward or by sticking AI viewers with a higher ticket cost. Which would be yet another reason for AI to kill us. Or perhaps the lower quality of the recording of the event (such as the coughing of the meatbag members of the audience) relative to a purchased recording would offset this.
If an AI had human-like memory and forgot stuff, then they could be treated as human consumers—since they would be analogous to humans in this regard. Another option is that that AI systems could be required to have a special app for “degrading” their memory of copyrighted media so that they would be analogous to humans in this one area. On the plus side, this would allow an AI to enjoy works repeatedly, on the downside they might consider this just another reason to kill all humans.
Reblogged this on newauthoronline and commented:
This is an interesting post, written by a philosopher. I really don’t think that authors need to concern themselves, in the foreseeable future (perhaps never) with the possibility that artificial intelligence will infringe their copyright. As for us placing restrictions on the ability of AI to infringe copyright (and this being “another reason” for them “to kill us”) I chuckled at the very idea and I suspect the author of this post had his tongue in his cheeck when he wrote those words.
Fahrenheit 451 Terminator!
Legally, the issues involved have already been well thrashed out. Several cases have already covered this ground. For example: https://en.wikipedia.org/wiki/Field_v._Google,_Inc.
The tenor of the decisions seems to be that copying for internal storage or transformation is in fact permitted, assuming the copying device/browser/AI has been granted access to the work, but reproduction is prohibited. The history of the case of Google Books is nasty, and in the end, Google won mostly by throwing their weight around, but since it is an article of faith for them that they are not Evil, I am sure they are entirely satisfied with their position.
So an AI could certainly learn from and enjoy copyrighted works, but would have legal exposure if it reproduced them.
This, incidentally, is the same situation as for humans. There is a small community of book copiers. They scan and OCR and type copyrighted books, and release them as files on Usenet and Discords and IRC. There have been several lawsuits. Releasing the copies is the act that exposes the copiers to liability.
Spoiler alert: Google kills us all.
Of course we can trust Google; after all evil people never say they are not evil.
I think in a larger sense, there is a lot more truth to this than you intend. Although “Google”, and by extension, “The Internet” are not necessarily inherently evil, information is a very dangerous thing in the wrong hands.
Kinda like “The Force”
Any form of power in the wrong hands is a dangerous thing. Including, some might say especially, the power of academia and especially the “credential system” on which much of it rides.
after all evil people never say they are not evil.,
As per Locke, one of my main duties is to preserve humanity.
As everyone knows, knowledge is power, France is bacon.
The common thread of almost all of my posts on this blog has been about the misuse of information, the proliferation of limited, or one-sided information, the mob-mentality that arises from the pervasive laziness that prevents people from doing their own critical thinking, and the obsession we now seem to have with opinions, politics, current events, the “scandal of the day” along with our zeal for posting and reposting limited raw information.
When I was about 10 years old, I said to my mother, “You know, mom, the ‘truth’ doesn’t seem to matter as much as what people think is true – it’s what people think is true that guides their actions.” She was appalled that I would think that – interpreting it as some internal justification for lying – but it really is quite true.
Academia has the power to open doors, but you are right – the “credential system” is deeply flawed. A degree in and of itself is meaningless. I have this argument with my students all the time – especially in my field, where no one looks at a GPA; it’s the portfolio, the problem solving ability, the creative and innovative thinking, and the ability to collaborate that are the most important. The worst feeling for me is when a student says, “I have my degree, I have a 3.5 GPA – why can’t I get a job?” The answer, of course, is the underlying reason that they think that’s a valid question.
One thing that I continue to see as a positive about the posts on this blog is that Mike, as a teacher, is really doing a great job with us – whether on purpose (I hope) or inadvertently. He has got us all thinking, arguing, posting contrary opinions and defending them, and doing at least primary, and in some cases secondary, research to bolster our points of view. Taking academia as a whole, the biggest issue for me is the “Ivory Tower” position – the “Sage on the Stage” attitude. My best teachers throughout my life have been the controversial ones – those who are not afraid to profess a strong point of view and who are willing to defend or back off that point of view in the face of contrary arguments or harsh criticism. It is my hope, as others have expressed, that when students are faced with complex, multifaceted and deeply nuanced issues they are not afraid to contest, to oppose, to argue, and to explore. Definitely “Pollyanna-esque”, but then there’s that whole leading a horse to water thing. Some people “get it”, others do not.
Going back to the original point about information and its use/misuse, I heard an anecdote not too long ago, posing the question, “If someone from the 1950’s were to travel forward in time to 2017, what do you think that person would find the most shocking/appalling?
The answer, “I carry a device in my pocket with more computing power than went to the moon in 1969, that connects me to all the knowledge in the world for all of history – and I use it to look at pictures of cats and to <a href="https://aphilosopher.wordpress.com/"argue with strangers ”
If only it were so innocuous.
re – last link … “oops”
Agree so much here I’m almost uncomfortable with doing so. Which is kinda the point of these things. When I was about 10 years old, I said to my mother, “You know, mom, the ‘truth’ doesn’t seem to matter as much as what people think is true – it’s what people think is true that guides their actions.” . Had the exact same thought at about the same age. Maybe 7th grade or so. Ran away from it, came back to it, has disturbed me ever since.
As for and I use it to look at pictures of cats …, whenever I see that quote I am reminded that some in the Roman Empire had harnessed the power of steam. They only seem to have used it to open temple doors and make toys. AIUI, the Incas had the wheel but it was only used as a child’s toy. Or such is what people think is true. I sometimes have my doubts about such facts.
“The tenor of the decisions seems to be that copying for internal storage or transformation is in fact permitted, assuming the copying device/browser/AI has been granted access to the work, but reproduction is prohibited.”
Agreed. The spirit of copyright laws is to protect the intellectual property of the creator, and (largely) to prevent others from reselling or otherwise inhibiting the creator’s ability to make a profit from their work.
The issue of “Fair Use” comes to mind, and the four-point due diligence test for it.
Looked at another way, one might consider that a creative work such as a work of art, an article, or a book, is meant to be intellectually transformative, contributing to the worlds of ideation, analysis, or just raw information. In that sense, “programming” an AI entity with scanned volumes absolutely serves that purpose, with the actual act of copying being nothing more than a means to an end.
You might TiVo a favorite show or football game to watch later with a group of friends – that act in itself violates the strict letter of the law, but serves the intent of the creators quite well. In fact, I think those copyright notices have been adjusted to account for this kind of copying. Maybe it’s time for the language of copyright laws to be revisited to account for next generation technology.