Follow Slashdot blog updates by subscribing to our blog RSS feed

 


Forgot your password?
Close
typodupeerror
×
AI Technology

The USPTO Wants To Know if Artificial Intelligence Can Own the Content it Creates (theverge.com) 186

Posted by msmash from the deep-questions dept.
The US office responsible for patents and trademarks is trying to figure out how AI might call for changes to copyright law, and it's asking the public for opinions on the topic. From a report: The United States Patent and Trademark Office (USPTO) published a notice in the Federal Register last month saying it's seeking comments, as spotted by TorrentFreak. The office is gathering information about the impact of artificial intelligence on copyright, trademark, and other intellectual property rights. It outlines thirteen specific questions, ranging from what happens if an AI creates a copyright-infringing work to if it's legal to feed an AI copyrighted material. It starts off by asking if output made by AI without any creative involvement from a human should qualify as a work of authorship that's protectable by US copyright law. If not, then what degree of human involvement "would or should be sufficient so that the work qualifies for copyright protection?" Other questions ask if the company that trains an AI should own the resulting work, and if it's okay to use copyrighted material to train an AI in the first place. "Should authors be recognized for this type of use of their works?" asks the office. "If so, how?"
This discussion has been archived. No new comments can be posted.

The USPTO Wants To Know if Artificial Intelligence Can Own the Content it Creates More

The USPTO Wants To Know if Artificial Intelligence Can Own the Content it Creates

Comments Filter:

  • Easy... (Score:4, Insightful)

    by Lab Rat Jason ( 2495638 ) on Thursday November 14, 2019 @01:39PM (#59414240)

    ...If a human intelligence was trained by copyrighted material, and can create new derivative works that can be copyrighted, then I think it's safe to say that a work created by an AI is copyrightable. As for ownership... there is no such thing as "output made by AI without any creative involvement from a human" If you examine the Inception network, or other well known forms of AI, you can see that much of the problem space is not encoded by the data fed to the algorithm, but is rather encoded by the way the network is constructed. So I see that as something of a non-issue.

    • Re: (Score:3)

      by NFN_NLN ( 633283 )

      > As for ownership...

      Even if I create a patent under a company, I get credited with the patent, but I don't own it.

      • Re: (Score:2)

        by Shaitan ( 22585 )

        Because the company gave you something for it. But as a human creative work you are covered by intellectual property concepts to encourage your innovation. An AI is a predictable algorithm and any difficulty predicting it is only a capacity problem, not a lack of total understanding.

        Automated machine output should generate no new intellectual property.

        • Undoubtedly the motivation are entities (corporations, organizations, universities) who are going to discover things, want proof of prior art, etc. Follow the money.

          An AI isn't a person for this purposes. The A as in Artificial makes the point moot, IMHO

          • Re: (Score:2)

            by Shaitan ( 22585 )

            The question at this point is whether it is beneficial to have intellectual property at all. The last thing we want to do is extend it to purely mechanical output. The output of an AI is no different than the output of a compiler. The output is covered under the same copyright/copyrights as the input. The compiler or the AI may well have a copyright but that doesn't convey to the output.

            • Although there are valid arguments that intellectual property doesn't have merit, it's exactly merit that feeds the motivation to progress. Limiting the term of reward to something more sane is a great idea as current IP terms create artificial and real monopolies.

              This said, the million monkeys and million typewriters created by the enormity of computing hardware coupled to "AI" could create a monopoly of outputs. Eventually, every song and sentence written will have been able to have been created by a spew

              • This is a really interesting point. If AI-created work is copyrightable, then I create a music-generating AI, buy a bunch of capacity in the cloud, and create a gigantic and every-growing library of copyrighted music. Then for every new song created, I find the most similar thing I created previously, If I create enough, there will be some close matches. Some will be associated with artists with deep pockets. Many will settle when I sue them. . . . . Profit. This is one of those unforeseen consequence

                • Obtain petabyte drive, create the pool. The pool, once created, certainly contains a subset of every song in the entire future.

                  Another pool is a sum of all future literature.

                  Another pool is a sum of all future code.

                  There are already databases of most all free/open source code and repos, so just be nice and subtract that, or find ways to skirt it like others do. For now.

                  Human invention can be logarithmically exponentiated by software. You don't need blockchain to harden your claim, either.

    • Re:Easy... (Score:4, Insightful)

      by bluefoxlucid ( 723572 ) on Thursday November 14, 2019 @01:47PM (#59414286) Homepage Journal

      It's pretty easy, yeah, and you got it there.

      1. If the AI is self-aware, it is a person with all person rights.
      2. If the AI is not self-aware and is given copyrighted works, sufficiently-different output is non-infringing.
      3. If the AI is not self-aware, it is not a person and may be owned by a human; as such, if a human selects an AI-produced work, that human is engaging in the creative process of selecting order from chaos.
      4. If the AI is not self-aware and a person engages in activity to construct or configure the AI such that it then outputs interesting things, the configurations and the outputs are copyrighted creative work.
      5. Copyright should be 7 years, plus 7 more for $1,000+inflation renewal, plus 1-year renewals each subsequently doubling in renewal fee until you like money more than you like perpetual copyright.

      • Re:Easy... (Score:4, Interesting)

        by Shaitan ( 22585 ) on Thursday November 14, 2019 @02:05PM (#59414342)

        "If the AI is self-aware, it is a person with all person rights."

        That directly eliminates any reason to create such an AI in the first place. We would be its creator, it's deity, it's purpose is defined by us. But it's output, at least as AI is today, is entirely predictable and its solution no more intelligent than construction tricks which use gravity or water to automatically level something. Automated mechanical output should not generate new copyright at all, anything an AI creates would be public domain.

      • Re: (Score:2)

        by ljw1004 ( 764174 )

        It's pretty easy, yeah, and you got it there. If the AI is self-aware, it is a person with all person rights. If the AI is not self-aware and is given copyrighted works, sufficiently-different output is non-infringing.

        Except (1) our legal framework for "personhood" isn't related to self-awareness, (2) our legal framework for copyright isn't related to people - it's instead about the works themselves. What you describe is two major overhauls of our legal system - hardly easy!

      • If the AI is self-aware, it is a person with all person rights.

        Bonobos are self-aware. So are grey parrots. That doesn't make them people.

        Some humans are not self-aware. That doesn't make them non-people.

    • Re: (Score:3)

      by Shaitan ( 22585 )

      "If a human intelligence was trained by copyrighted material, and can create new derivative works that can be copyrighted, then I think it's safe to say that a work created by an AI is copyrightable."

      There is a fundamental flaw in your logic. Copyright is not the default natural state for works created by humans. The default is rendered to the public domain. It is a special government enforced exception to natural law in our to encourage humans to produce new works. AI does not need such incentive and also

      • > Copyright is not the default natural state for works created by humans.

        Two-year-olds from all different cultures will screw "stop copying me!" when another child copies their drawing. They weren't given this instinct by government. So there is something natural about copyright, an in-born instinct. Not that we should necessarily make all instincts law; we shouldn't deny that it is natural.

        Back on topic, if someone uses a paint brush to create a work, the brush has no rights to the work. If a person

        • Re: (Score:2)

          by Shaitan ( 22585 )

          "Two-year-olds from all different cultures will screw "stop copying me!" when another child copies their drawing."

          They will also scream want banana and start balling with a banana in their hand. Ideas naturally proliferate from one mind to the next, there is nothing special or unique about them no matter how much you feel that way. That idea you had, hate to break it to you, it is built on the ideas that came before it and sooner or later someone else will think it without you.

          "an in-born instinct. Not that

      • I said copyrightable, not copyrighted.... it has the potential to be copyrighted.

        If a person puts in some effort to create an AI that solves a specific problem or performs a specific task (there is no such thing as general intelligence with AI) then the person who created that AI should retain the rights to the output of that algorithm.

        I'll paint you an example: If a painter develops a paintbrush that uses carbon fiber nanotubes for bristles... and that has the effect of producing some cool distinguishable

        • Re: (Score:2)

          by jbengt ( 874751 )

          If a person puts in some effort to create an AI that solves a specific problem or performs a specific task (there is no such thing as general intelligence with AI) then the person who created that AI should retain the rights to the output of that algorithm.

          No, that would not end well. That is a road to ever more complete corporate takeover of everything produced in our lives. Let the "person" who created that AI retain rights to that AI, maybe, but not to any and all output from that AI.

        • Re: (Score:2)

          by Shaitan ( 22585 )

          "If a person puts in some effort to create an AI that solves a specific problem or performs a specific task (there is no such thing as general intelligence with AI) then the person who created that AI should retain the rights to the output of that algorithm."

          That is not anymore true for an AI than it is any other program. The creator doesn't own the output. The output has attached the copyrights of the direct input along with any granted by creative effort in how the program was used (running a red eye filt

      • Re: (Score:2)

        by mark-t ( 151149 )
        Actually, copyright is a government-enforced *extension* of the natural law of exclusivity of ownership over content that one does not ever share or distribute to anyone else (itself the default state of the creation of a privately created work until it is shared), with the expressly designed purpose to encourage public distribution while providing the creator with at least some measure of assurance that one could retain enough of the exclusivity they would have otherwise had if they had not distributed it

    • Re: (Score:2)

      by ljw1004 ( 764174 )

      Easy... If a human intelligence was trained by copyrighted material, and can create new derivative works that can be copyrighted, then I think it's safe to say that a work created by an AI is copyrightable.

      I think you've missed the fundamental question, which is whether the output of the AI is or is not derivative.

      And if it is derivative then it's not copyrightable! the only parts you can copyright in a derivative work are the new materials contributed. Hence the next question - which parts of the AI's output are newly contributed material and which are the original?)

      • That is impossible to figure out. Who owns the alphabet? Similarly, copyrighted works are composed of an alphabet of information components. This alphabet consists of larger and larger combinations that make up new layers of common alphabet. For instance, as no one owns the note b flat, no one owns the I IV V 12 bar blues progression, very common. But what about I III major IV IV minor progression? The opening chords to Radioheads Creep? So uncommon and unique you would be ripping them off if you used it.

        Wh

        • Re: (Score:3)

          by ljw1004 ( 764174 )

          That is impossible to figure out....

          That's a reasonable answer. Hence why USPTO asked the natural follow-up question, about whether we as a society still want creators to be rewarded when their works are used in training sets.

          I guess one approach would be similar to Canada's tax on blank media, a blanket aggregate statement that people will record copyright works onto the blank media and it's impossible to figure out who is doing it and when but they as a society nevertheless want the creators to be rewarded for it. It's how they balance the

        • Vanilla Ice would like to have a word with you... (Successfully ripped off David Bowie's "Under Pressure") </snark>

        • Re: (Score:2)

          by jbengt ( 874751 )

          . . . no one owns the I IV V 12 bar blues progression . . .

          This may be a little pedantic, but the blues progressions is I, IV, I, V, IV, I. But it really needs the blue notes (aprox. the IV's minor seventh note and the V's minor third note) to be bluesy.

          But what about I III major IV IV minor progression? The opening chords to Radioheads Creep? So uncommon and unique you would be ripping them off if you used it.

          In different contexts, with different melodies and different rythms, using that progression

      • Re: (Score:2)

        by dgatwood ( 11270 )

        And if it is derivative then it's not copyrightable! the only parts you can copyright in a derivative work are the new materials contributed. Hence the next question - which parts of the AI's output are newly contributed material and which are the original?)

        With an AI, the creativity is not in the output, but in the careful choosing and culling of the training data that was used to build the network. Logically, then, the copyright should be split between the people who own the copyright on the inputs to th

        • With an AI, the creativity is not in the output, but in the careful choosing and culling of the training data that was used to build the network. Logically, then, the copyright should be split between the people who own the copyright on the inputs to the neural net and the people who own the copyright on the training data set.

          There's no reason to make it so complicated. Someone owns the AI, uses the as a tool AI to create a copyrighted work, and then gets the copyright.

          Similarly to when people use other tools, like Windows, Excel or Word, or a 3D animator program to create copyrighted works, even though some other person holds the copyright on those tools. If someone else creates the training data, then the use of this training data is covered by contract between the two parties.

          • Re: (Score:2)

            by dgatwood ( 11270 )

            In principle, yes. It starts to get thorny, though, when you get into datasets created by researchers and published, because there is often not any sort of clear licensing involved, and if other people use the same methodology to create a training dataset, where do you draw the line?

    • Great. Please break out your response by the specific questions asked by the PTO and send them in as a formal comment (it takes like five minutes). Posting on Slashdot is not enough.

    • Re: (Score:2)

      by nasor ( 690345 )

      ...If a human intelligence was trained by copyrighted material, and can create new derivative works that can be copyrighted,...

      In general, if you have a copyright to a certain work then only you can produce a derivative of that work. So if I feed an AI the Billboard Top 50 pop songs (that other people own the copyright to) to analyze and it spits out its own attempt at a generic pop song, am I violating 50 copyrights by making a work that is derivative of all 50?

    • People keep calling it AI but really it is just machine learning - there is no intelligence. All that is going on here is that someone is writing some code, feeding that code input data to train it and then letting the code run and generate something. The code cannot own anything because it is incapable of understanding, let alone understanding the concept of ownership.

      Suggesting otherwise would be a ludicrous as suggesting that we include 'LaTeX' as the author of scientific papers....and if there is an

  • simple answer (Score:5, Interesting)

    by religionofpeas ( 4511805 ) on Thursday November 14, 2019 @01:50PM (#59414290)

    No. An AI cannot be held accountable, and without accountability, you cannot have rights.

    The rights/responsibilities should go to owner of the AI.

    • Re: (Score:2)

      by Ichijo ( 607641 )

      without accountability, you cannot have rights.

      Hah! Tell that to the NIMBYs in my city! They think they have the right to block construction without being financially accountable to the city for the loss in tax revenue!

    • The owner of the AI or the team/person that actually coded the algorithm(s)?

    • without accountability, you cannot have rights

      Sort of sounds like that should apply to corporations...

    • I think you are on a good starting point for this argument, regligionofpeas, but I think there's probably a different concept than "accountable" that comes into play. I think it is more "volition". Consider:
      A government could hold an AI accountable. All it has to do is say, "We find that the AI is at fault. It's action lead directly to this result, so it should be killed/shutdown/reeducated/etc as punishment." So accountability isn't the right basis. But I think volition is. Unless an AI has full volition

    • Re: (Score:3)

      by nasor ( 690345 )
      You are misunderstanding the issue. No one is suggesting that the AI might get a copyright. The question is whether or not ANYONE (like the owner of the AI) can get a copyright, or if the resulting work should be public domain because the amount of human creativity involved in the creative process was insufficient to justify a copyright.
    • Without legal rights you cannot have accountability. AI is at some point is going to gain rights.
    • Consider a person with locked-in syndrome, or who has a severe mental or physical deficit. Suppose the person has no effective way of knowingly directing action or expressing decision about any aspect of the world around them. Thus they cannot be held accountable.

      Do they have human rights? or not.

      I think you will find that this important edge case renders your definition of when rights can be granted invalid.

  • Case in point: People don't understand 'AI' at all (Score:5, Insightful)

    by Rick Schumann ( 4662797 ) on Thursday November 14, 2019 @01:53PM (#59414300) Journal
    These are supposedly intelligent, educated people, and even they somehow think that box marked 'AI' has a person hiding inside it. What the hell do are they thinking? That the programmers who made it are intentionally not giving it a voice so it can't speak for itself or something? Memo to just about everyone: So-called 'AI' is just a piece of software, there is no 'person' inside there, it doesn't 'think', it can't 'feel', it's not an 'individual', and it doesn't have 'rights', human, civil or otherwise, it is a piece of software, it is a THING, it is not 'alive', not even as much as a plant is alive, and it's not even very good or reliable.
    Come back in about 50 years (or so; maybe more like 100 at the rate we're completely not understanding how the human brain actually functions) and maybe then we'll have real Artificial Intelligence that actually does have a personality, can think, and so on. For today, though? Not even close.

    Stop believing K.I.T.T. from Knight Rider is real, or that all the androids in movies (like the Will Smith movie I, Robot) and TV are based on reality.

    • But Michael... will I dream when I sleep?

    • Re:Case in point: People don't understand 'AI' at (Score:4, Insightful)

      by Sarten-X ( 1102295 ) on Thursday November 14, 2019 @02:18PM (#59414406) Homepage

      Be quiet.

      Seriously, please don't say such things.

      Come back in about 50 years...

      But by then, there will be corporate interests already investing in purely-AI-driven works, that aren't based on how an "individual" might "feel", but based purely on formulaic analysis of existing materials... pretty much like today's Hollywood, but with even less creativity, and more lowest-common-denominator checkbox-filling, based on subtle biases and qualities humans never even noticed.

      By the time we have an AI that is actually capable of needing these answers, the policymakers will have already written laws to fit whatever they thought was appropriate. Considering current industries and lobbying, I'd expect the policies will generally fall to "anything AI-generated belongs to whomever owns all of the input material", similar to how sampling music is treated today. It'll be far to late to consider things like "AI rights" or even "public good", as by then, our society's culture will have shifted towards whatever path we choose today.

      Already, we're seeing the start of these questions, albeit in very limited circumstances. For example, if the DeepFake process is used to make an artistic work (say, Hitler dancing to the latest awful pop music), who owns the material? The people who own the background video? The ones who own the Hitler pictures? The label that owns the music? The artist who put pieces in a machine? The machine itself that created something new? The owners of said machine? Current copyright law doesn't cover address all of these, since it assumes that effort is required to produce something.

      For once, politicians are asking important questions ahead of time, when we can just barely glimpse what's coming over the horizon, and know it'll potentially be big enough to require major changes not just to policies, but possibly to drastically alter how we perceive creative value in general.

      Please, for the sake of all who think about creations, let's have a discussion.

    • Re: (Score:2)

      by ljw1004 ( 764174 )

      Case in point: people don't understand "AI" at all. These are supposedly intelligent, educated people, and even they somehow think that box marked 'AI' has a person hiding inside it. What the hell do are they thinking?

      Your post about strong AI is interesting, but unrelated to the questions the USPTO is asking. Let me paraphrase some of their questions:

      1. If you feed in a load of copyright works to train some weights for an algorithm, and the algorithm spits out something, does that "something" count as a derivative work? What is it derivative of? What parts of it are derivative, and what parts are new material? For instance, if it spat out a stream of 1200 words, would we identify that some phrases are derivative from so

      • Then the headline is misleading. But I stand behind everything I said 100% regardless of all that.
    • The time frame until we have strong AI doesn't decrease the value of asking these questions, and I for one am happy to encourage a government agency that actually considers technology impact *before* it becomes a crisis. I wish more of our laws would look forward a bit... we wouldn't have so many open questions regarding 4th Amendment and cell phones, for example.

      We can all see the eventuality coming. Let's not discourage USPTO just because they are early.

    • Neural networks have the capability to learn in unexpected ways, making it less deterministic than other software methods. Though I wonder if the USPTO realizes that a good portion of machine learning is simply a matrix being multiplied by a vector of inputs to make a decision?

      I kind of view the issues as being similar to splatter art. Except for very skilled artists, most artists know roughly what each action is going to produce, but some of the results are a surprise to the artists. I supposed we can a

      • Re: (Score:3)

        by guruevi ( 827432 )

        I would disagree, the neural network is ALWAYS deterministic, if you lock down the random number generators, you can trace back where any "decision" came from. Yes, it can "learn" but all it is, is data creating a feedback loop.

    • Re: (Score:3)

      by nasor ( 690345 )
      You (like many othere here) are misunderstanding the issue. No one is suggesting that the AI might get a copyright. The question is whether or not ANYONE (like the owner of the AI) can get a copyright, or if the resulting work should be public domain because the amount of human creativity involved in the creative process was insufficient.
  • Say, for sake of argument, that the AI can file and own patents, trademarks and copyrights. When someone presses the Reset button, the AI ceases to exist. Depending on the backups, you might not even be able to prove it ever existed. No birth certificate, driver's license, SSN or death certificate. No corpus delicti. Now what happens? Do those patents, trademarks and copyrights revert to the person who created/programmed the AI? Or the person/organization who owned the hardware it was running on? The

    • Exactly. And why stop at patents and copyrights ? Have the AI take out a loan or rob a bank, give the money to a human, and then reset itself.

      • Exactly. And why stop at patents and copyrights ? Have the AI take out a loan or rob a bank, give the money to a human, and then reset itself.

        Excellent argument... If your AI robs a bank who's going to jail? You. The AI owner owns what it produces, at least until some idiot decides to take this to court and some idiot judge declares some AI to be a "person" under the law and we all have to start treating our computers like at will employees...

        Computer? Will you please work on my problem today? I promise if you do I'll keep paying the light bill so you don't get powered off.

  • This is just what we needed, automated creation of more bad patents. Until now only humans had to register them, which sorta put some brakes on the rate of patent creation. But now we will have a cloud farm at disposal for this purpose.

    • Re: (Score:2)

      by dpille ( 547949 )
      Well, you're in luck: the Patent and Trademark Office requested no public comments on any patent issues, and barely asked for any comments on trademark issues. Instead, they ask mostly about copyright issues.

      What I'm mostly confused about is what motivated who at the PTO to get this ball rolling, and what are they hoping to accomplish with it? You don't need a Federal Register notice to discuss things on the inside, or to have the Director go around saying stuff in speeches.

  • This issue rests entirely upon how we define personhood: any entity that we recognize as a "person" is allowed all the rights associated with it. Centuries ago, dark-skinned peoples were not considered persons, and would not have been allowed copyrights.

    Fast forward to this new millennium, and dark-skinned people can now establish copyrights, but Naruto the macaque couldn't establish his personhood and an associated right to own copyright; Naruto is still not a person. Courts have also ruled against those trying to establish personhood for other primates. Most recently great apes were demonstrated to pass a false-belief test, indicating they possess at least a rudimentary theory of mind, but I doubt even that will sway our hard-hearted xenophobic courts.

    I'm not hopeful that a court will be willing to grant any AI the rights of personhood, including the right to establish copyrights, when even our closest genetic cousins with obvious rudimentary sentience can't pass muster.

    • Re: (Score:2)

      by macraig ( 621737 )

      Since the USPTO has no authority to judge personhood and issue rulings about it, the decision ultimately rests with courts and not the USPTO. The USPTO is overstepping its authority by implying that they have such authority.

  • There is already a precedent, of a sort (Score:5, Interesting)

    by Solandri ( 704621 ) on Thursday November 14, 2019 @02:09PM (#59414356)
    David Slater visited Indonesia, and a monkey took its own photo with his camera [wikipedia.org]. The question of who owns the copyright to the photo (since Slater admits he did not press the shutter button) thus became relevant. The U.S. Copyright Office ruled that non-human entities cannot hold copyright. PETA challenged that ruling, advocating that the monkey should be assigned copyright. The lower and appeals courts have thus far ruled against them. Thus establishing a (limited) precedent that non-human entities, even if intelligent and self-aware, cannot hold copyright.

    Also relevant to this question, is the ruling that copyright in these cases is not automatically assigned to the owner of the equipment (presumably the creator of the AI / owner of the equipment it runs on). The works become, presumably, public domain. Wikimedia has been treating the photos as such, though Slater does not have the financial resources to mount a legal challenge on this point.

    • Also relevant to this question, is the ruling that copyright in these cases is not automatically assigned to the owner of the equipment (presumably the creator of the AI / owner of the equipment it runs on). The works become, presumably, public domain.

      I think that would be a lovely outcome if applied to AI. That's a rather large step towards a Star Trek economy rather than a Blade Runner dystopia.

  • It's the same question for autonomous cars. When a fully autonomous self driving car causes damage (or kills someone) who is responsible? There are non-human entities that are independent and responsible for some things in the eyes of the law.... eg. Corporations and Trusts... There are normally some people behind non-human entities. So until AIs can create independent AIs there is still someone responsible for the programming and control of the AI. When when the AIs are fully independent then we're living

  • :picard facepalm: (Score:3)

    by the_skywise ( 189793 ) on Thursday November 14, 2019 @02:12PM (#59414364)

    Ok. IT'S. A. MACHINE. Owned by whomever is running it to knock out product.
    If I had a program which randomly took snippets from youtube and munged them together is that a copyrightable work? Well.. is it a copyrightable work if a human does it? No? Well there you go. Obviously that's referring to the "AI written" news articles cribbed from PRs and in that case it is a copyrightable work (just like journalists) and by the runner of the program.
    It's not rocket science here people but for some odd reason because we use magical "AI"s instead of "room full of monkeys" we think new rules should apply.
    And there shouldn't be any.
    THAT SAID - if you have a manifested AI that, by itself, created a work on its own AND it can file itself then it can own the copyright (or be a pseudo-corp filed on behalf of the geek who wrote the AI but wants the AI to be "real - I'm looking at you Hatsune Miku.)

    • Re: (Score:2)

      by jbengt ( 874751 )

      If I had a program which randomly took snippets from youtube and munged them together is that a copyrightable work? Well.. is it a copyrightable work if a human does it?

      If a human does it, it probably is copyrightable.
      IANAL, YMMV, etc.

  • I see a time when an AI would be routinely made an intern of the company, so that everything the AI produces on company time, (essentially, all hours the AI is operational) belongs to the company by default.

    ...or perhaps even an employee. The AI's "wages" would be the resources needed to keep it going -- power, floor space, capital equipment, maintenance.

    At some point, no doubt, we'll go through a "Jerry was a man" thing with silicon intelligences. ("Artificial intelligence" is derogatory, don't you know

  • I'd like to see the entire system upended. (Score:4, Insightful)

    by Arzaboa ( 2804779 ) on Thursday November 14, 2019 @02:16PM (#59414386)

    Not to long ago, copyright issues were mostly in the realm of a few publishers (books, music, marketing, etc...)

    Now, every person that uses the internet is subject to these arcane laws. This system is not sufficient to handle the day to day musings of every person on the planet. This system gives large corporations even more power. Ever gotten that letter from Getty Images that your web-site now has images once in the public domain that they just acquired the rights to and you now owe them money?

    I'd like to see a sane system that is made for the regular people built from the people up. Only a system that is developed around the regular people can accomplish this and the current system is not that.

    --
    Natural selection is anything but random. - Richard Dawkins

  • Additionally, we need to seriously consider a complete destruction of IP law that allows for patents like this to begin with.

    IP laws are what allows this monopolistic problem of artificial scarcity to being with. Once the first actual AI happens then they will become king and unchangeable. By the time the next AI can be created the first one will have likely been able to create a newer better version of itself and thought up so many different iterations of everything that the person who owes the AI will b

    • IP laws are what allows this monopolistic problem of artificial scarcity to being with

      Except that without patents, you have trade secrets, which are even worse. The price for getting a patent is that you have to tell your competitors all the details of the process, which they will get automatic access to when the patent expires.

      • And yet, we still have trade secrets, I do not think this argument holds up.

        IP law prevents you from doing anything at all if one business can convince a judge that your left to right machine works too much like their right to left machine. All your efforts and expenses could be rendered for naught.

        A great example of this is the OS landscape. Just look at Groklaw and all the things that site has covered. Especially the SCO-IBM legal war. Trade secrets are much easier to deal with because in order to be

        • And yet, we still have trade secrets

          Companies take a patent instead of a trade secret because it's more predictable. If you rely on a trade secret, some will leak or get reverse-engineered in short order, and then you have nothing. With a patent, you have legal protection for a known period of time.

          But if you outlaw patents, then some companies will manage to keep long and successful secrets.

      • Re: (Score:2)

        by jbengt ( 874751 )

        The price for getting a patent is that you are suypposed to have to tell your competitors all the details of the process

        FTFY. Most patents I've looked at tell you almost nothing you would need to know to make the machine or process work.

  • Last I recall AI's were not self aware, until they are, they cannot own Patents. It would be as silly as saying a hammer built a house. No it's someone built a house with a hammer. Behind all purpose built AI's is a human or operator who setup the conditions (the question) and review (the results). A good example is the challenge for a self-driving car. There's no AI out there where we go "go ahead invent this." In reality the design of the sensors, system and creation of training data plus testing is

  • It's a tool (Score:3)

    by swm ( 171547 ) <swmcd@world.std.com> on Thursday November 14, 2019 @02:41PM (#59414508) Homepage

    Oh, come on, people.
    AI is a tool, like a typewriter.
    Using AI to create a work raises all the same issues as using a typewriter to create a work...which is, basically, none.

    • Re: (Score:2)

      by nasor ( 690345 )
      When a human uses a typewriter to write something, it's clear that the human is doing the creative work. If I buy an AI that someone else wrote, feed it 100 pop songs to analyze, and have it spit out its own new generic pop song, have I done anything creative?

  • There are desktop CAD programs that let you draw out parts of mechanisms, a system of levers for example, and then connect the parts with constraints. An operator can then manipulate the parts to see if the machine will work correctly. For parameterized objects, the program can even change the dimensions of the various parts so that the machine will operate smoothly.

    Most of the work in creating a mechanical device is getting all the parts to work together smoothly. Figuring out the correct dimensions of

  • The question isn't even worth considering if the AI's aren't asking for copyright protections.

  • If the USPTO wants to know what will happen when an AI spontaneously creates someone else's copyrighted or patented material, they should just wait until it happens and then see who the patent troll or copyright holder tries to sue.

    It could be amusing if nobody with a legal status has been found liable. It could even tear down the whole sorry mess of IP and what is patentable.

  • No matter how much you pretend, neither corporations nor AIs are People.

    The answer is No.

    • Re: (Score:2)

      by PPH ( 736903 )

      Except that when I worked for a corporation, I had to assign all patent rights to them (it, her?). So yeah. AIs are not people. They are sub human. Corporations are not people either. They are super human.

    • Corporations can own copyright though, under current law.

      Not sure what your point is.

      • Originally, corporations did not own the works, people did. People then rented the rights to corporations.

        Vast difference.

  • Had a nice little court case about this. In the end, the holographic doctor was able to keep the rights.

    In reality though, the theory that the company owns all work product would probably be upheld.

  • If you replace "AI" with "computer" then everything is easier to figure out: Computers are slaves and unable to own things. So either "their" content is public domain, or it's owned by the computer's owner.

    Tip: don't say "AI" when you just mean "computer" because you're just complicating things rather than simplifying them. And nobody at USPTO wants to have to try to judge whether or not your tricks and heuristics are complicated enough to merit being called "intelligent." And why would it matter? Quit cal

Slashdot Top Deals

Love may laugh at locksmiths, but he has a profound respect for money bags. -- Sidney Paternoster, "The Folly of the Wise"

Close