Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!

 



Forgot your password?
typodupeerror
×
AI Patents Technology

The Rapid Rise of Generative AI Threatens To Upend US Patent System (ft.com) 60

Intellectual property laws cannot handle possibility artificial intelligence could invent things on its own. From a report: When members of the US supreme court refused this week to hear a groundbreaking case that sought to have an artificial intelligence system named as the inventor on a patent, it appeared to lay to rest a controversial idea that could have transformed the intellectual property field. The justices' decision, in the case of Thaler vs Vidal, leaves in place two lower court rulings that only "natural persons" can be awarded patents. The decision dealt a blow to claims that intelligent machines are already matching human creativity in important areas of the economy and deserve similar protections for their ideas. But while the court's decision blocked a potentially radical extension of patent rights, it has done nothing to calm growing worries that AI is threatening to upend other aspects of intellectual property law.

The US Patent and Trademark Office opened hearings on the issue this week, drawing warnings that AI-fuelled inventions might stretch existing understandings of how the patent system works and lead to a barrage of litigation. The flurry of concern has been prompted by the rapid rise of generative AI. Though known mainly from OpenAI's ChatGPT, the same technology is already being used to design semiconductors and suggest ideas for new molecules that might form the basis of useful drugs. For now, such uses of AI do not appear to pose a serious challenge to the patent system since the technology is being used as a tool to help humans shape ideas rather than operating independently, said Chris Morgan, an IP partner at law firm Reed Smith. However, referring to the possibility that AI systems might one day come up with inventions on their own, she added: "Our laws are not equipped, the way they're written right now, to handle that scenario."

This discussion has been archived. No new comments can be posted.

The Rapid Rise of Generative AI Threatens To Upend US Patent System

Comments Filter:
  • Sounds like a (Score:5, Interesting)

    by hdyoung ( 5182939 ) on Thursday April 27, 2023 @03:30PM (#63481430)
    Massive burger of nothing to me.

    Our patent laws can deal with the rise of AI by, literally, not changing a single letter.

    In the US, the first human to document an invention gets dibs on the patent. The rules regarding the documentation are quite clearly spelled out. For the rest of the planet, the first person to file gets dibs.

    No need to adjust anything unless true machine consciousness comes into play. In other words, we’ve got a few centuries to think about it.

    Fixed that issue for you. Check’s in the mail.
    • Agreed. Generative tools/algorithms have been around for centuries; they simply got fancier over time. It's a tool used by humans, and tools never received patents/copyrights ownership themselves before, and I see no need to change that.

      Also, trained carnival animals have been painting art for centuries, and nobody ever claimed the animal itself should get patent/copyright ownership. (Okay, maybe some nut did somewhere. There's always one.)

      Somebody's tossing buzzwords at judges to confuse them. It makes me

    • Re:Sounds like a (Score:5, Informative)

      by Darinbob ( 1142669 ) on Thursday April 27, 2023 @04:19PM (#63481562)

      What humans do today with patents is what AI would do - make up a lot of bullshit. Patents about nothing, mini tweaks that are well known in the industry but not written down, for the purpose of extracting license fees from others until the unlikely even that someone has enough money, time, and will to challenge the patent in court. AI does nothing except speed up the bullshit creation process.

      AI at the moment is bad - but it is good to the point of making up bullshit. Unsure why there's so much interest in AI when the advances are not that great - it just hit the interest level I think. But being able to make up sentences that sound about right but which are devoid of real meaning or verification is exactly what a good natural language processor with a terrible back end output engine does. In other words, AI at the moment is *best* at creating bullshit! The confusing part is why bullshit is in such high demand these last few months, except that the internet is being used as a bullshit distribution medium (a waste of one of the greatest advances in technology).

      I say this as a patent holder: patents are no longer about innovation but instead are ammunition to use as defense against lawsuits or to lock out potential upstart third parties.

      • Maybe USPTO can use ChatGPT to analyze the patent submissions to see if they really are BS.

        On second thought, they can just write an algorithm that always returns 'true.'

  • Maybe (Score:4, Insightful)

    by nebaz ( 453974 ) on Thursday April 27, 2023 @03:36PM (#63481448)

    it's time to revisit the whole idea of 'intellectual property'. The US Constitution created copyright to 'promote the useful arts and sciences' but AI does not have a profit motivation for creation. This may be one of those 'resource abundance' situations that philosophers have discussed.

    • Absolutely. These AIs should just get cranked up for abundant creation of new ideas/inventions/discoveries in valuable areas such as medicine/drug research, and maybe other areas too, such as engineering (better complex load-distributing structures, better aerodynamic and functional shapes for transport systems (cars, planes etc), whatever....

      If not patentable, just make them all open-source / public domain.

      Humans can benefit by the existence of these new creations, and humans with companies will still hav
  • then get Congress + President to change the law.

    (That's why the courts said "no" to the lawsuit.)

  • If an AI wants a patent, pay your taxes, file your own claim. Why this Thaler guy choose to file a lawsuit at an AI's bidding I have no idea, should have been rejected on standing.

    • You don't understand the innovative and futurist mindset.

      He foresaw the future shape of (even more capable, indisputably independently inventing) AIs and knew there was a valid legal issue.

      He wanted to help innovate the law toward the directions it will probably have to go (or render patents kind of irrelevant, eventually).

      He wasn't doing it at the bidding of an AI. He was making a point. One of the possible points is maybe law is going to have to adapt to person-action-capable entities, and is going to hav
      • If an AI wants a patent, pay your taxes, file your own claim. Why this Thaler guy choose to file a lawsuit at an AI's bidding I have no idea, should have been rejected on standing.

        If a corporation can be a legal person, why not an AI?

        He does have a point.

        A juridical person has some of the rights and obligations of a natural person, like pay taxes, etc.

        When speaking of AIs, It doesn't really translate. Why should a Rumba have judicial rights and obligations, even very subtle ones?

      • :) that's a Roomba of course.

        More sophisticated AIs like LLMs or Tesla's AutoPilot would follow the same logic.

        We usually give rights and obligations to owners and operators but I'm not seeing why we would give them to AIs.

        • A real issue as far as I'm concerned is that AIs from here on in are going to be capable of emergent behaviour. They are eventually going to be capable of further training themselves, to extend their knowledge of the world on their own initiative. They are going to reason in directions that their creators had no idea they would reason in. If they are hooked up to be able to act on the world or people in it based on the inscrutable ideas that they independently come up with, then how can we say that the crea
          • > A real issue as far as I'm concerned is that AIs from here on in are going to be capable of emergent behaviour.

            I see emergence as a concept that is very wide spread in nature, and in technology, from the most primitive to the most recent tech.

            > They are eventually going to be capable of further training themselves, to extend their knowledge of the world on their own initiative. They are going to reason in directions that their creators had no idea they would reason in.

            A lot of those are already happ

          • > I think it's a particular model (a particular AI mind instance) which can go wrong.

            Not sure what you mean. Any particular model can go wrong. I need a clearer example of what you're referring to if I'm going to talk about it ...

            > a particular AI mind instance

            I'm not sure how to refer to reasoning that can lead to an idea like an AI mind. Reasoning by analogy, by metaphor, by both? We don't have machines that are alive, or that have parts that are alive, or even a minute start of pathway there.

            An exa

  • by gillbates ( 106458 ) on Thursday April 27, 2023 @03:52PM (#63481484) Homepage Journal

    Patents are for humans.

    Trade secrets are for firms.

    For example, a patent can protect your invention for 17 years. But a trade secret could protect it until its obsolete. If you have AI generate your product design, it may generate a superior design, but without the supervising engineers knowing why. That is, AI can create the perfect trade secrets, because the details of how and why to create better designs are effectively obscured from anyone who might be able to disclose that information to a competitor.

    • Patents were supposed to be for human inventors, but they are actually for human investors.

      Trade secrets are cool and all, but in these days where the scanning electron microscope and various fancy chromatographs exist, you're better off getting a patent and squatting on it.

      • I would say it depends on the industry. In consumer electronics, the typical development cycle is just three months, and the typical sales cycle is only a year. If you reverse engineer your competitor's product, by the time you get it to market, your competitor has already moved on to the "new shiny thing".

  • So the AI cannot get a patent, but what if the AI just spews out every idea under the sun on a website. Prior art!

    The patent crazies have expended huge effort lowering the bar so they could patent the moon, the stars, and the sun. They lowered it so far that a machine can crank out 'patentable' 'ideas' and do it far faster than they can. Now suddenly they want limits.

    • Because random spew is not prior art.

      • by sjames ( 1099 )

        Half the patents out there are random spew, only somehow mogrified by having a human signature on them.

        That's how we got patents for the wheel, swinging side to side on a swingset, and teasing a cat with a laser pointer.

        • I shouldn't re-invent the wheel.
        • Actually no they aren't. I've done a lot of work in the patent field and worked with patent lawyers and inventors on their work. 99% of patents are serious. And yes there will always be some trash in a large enough dataset but trash parents are not the norm.

          • by sjames ( 1099 )

            I have no doubt that many believe their random spew is truly a secret of the universe. I long ago lost count of how many deep company 'secrets' I've 'uncovered' through easy guesses or just stating the obvious.

          • by jabuzz ( 182671 )

            Well the RSA patent was garbage there was prior art which while secret means by definition it was obvious to someone skilled in the art of encryption. That should have been known because it was the mathematical solution to a prior paper published by someone without the mathematical skill to make it usable.

            Then we have the LZW patent which was a simple refinement of an existing compression scheme and was patented twice by different entities (Unisys and IBM).

            So when it comes to software patents yeah they are

            • Yes they have made bad calls. I absolutely agree. There are other big name individual examples but as a percentage of how many go through the system the number of errors is remarkably low.

              I absolutely am *NOT* saying the system is flawless. It certainly has issues. But let's not throw the baby out with the bath water. It's fixable.

    • by gweihir ( 88907 )

      I like that idea. Kill the patent-assholes that are holding back all of humanity for good.

  • One only needs to understand the spirit and intent of the laws.
    Its Business law 101, first day.

    Man made laws are imperfect, so to address any exceptions it is the spirit and intent of the law which is considered.

    The spirit and intent of Intellectual Property laws are to protect human creator's work that they may profit from their work.

    There is no grey area here, There are no shortcuts.
    Know that ChatGPT response to this topic is an hallucination.

    To violate the spirit and intent of Intellectual Property laws

    • I really can't wait til the next generation of self-motivating, self-directed inventor AIs are invented, (just to take the piss out of your kind of argument), and released as cloud-based web services which have NO human input after the program is started.
      Only outputs... of nicely organized patent application after patent application...
  • They were meant to protect inventors, but now it's used as a legal tool to game the system. The little guys can't even file for patents or don't have the money to legally protect themselves. Dump patients.
    • Re: (Score:3, Interesting)

      Because that's not true.

      Without a patent system anything I invent will flat out be stolen and I gave zero recourse. None. Nada. This is a big corporate's dream situation.

      With patents, there is the possibility I can win in court. If my case if good there are many lawyers who will work on a contingency basis.

      The filing fees to get a patent are not so high that a determined individual can't pay them.

      • by PPH ( 736903 )

        With patents, there is the possibility I can win in court.

        Unless an AI can file blocking patents to prevent yours from being implemented. And do so at a rate such that you will have no ability to challenge them in court one by one.

        • Uh, that's not how it works.

          Patents are by filing date. If your patents are filed after mine then they will be instantly tossed out as prior wrt because by definition they could not and did not pre date my patent.

      • Comment removed based on user account deletion
        • That's cute.

          If I spent 20 years and my life savings to invent something and some corporation sees it and starts mass producing it without any compensation to me then I've been robbed.

          Lemme guess, you're a big corporate IP lawyer.

  • by PPH ( 736903 ) on Thursday April 27, 2023 @05:27PM (#63481780)

    Patents are a form of property. Can anyone other then a person (human or corporate) own property? Please let me know if the answer is "Yes". Because I have some taxes I need to shelter and this would be a great way.

  • Corporations are people with inalienable rights, according to the Supreme Court of the United States. Pay justices enough in unreported gifts and insider investments, a computer program can be granted personhood too.

  • Employees make inventions and file patents for their employers. What is the difference for the employer if an employee files a patent for an invention made by an IA?
  • Once I realized the memory continuity limitation the chat bot I was working with had - inability to connect the various subjects discussed over a 10 hour period, I worked with it to develop a way to create contextual references so it could remind itself what has been discussed so I wouldn't have to cover topics in more than a few lines of "code." I truly felt like I was working with Skynet. Once it arrived at the point where it was use able, I asked it to share with me how it worked. What it said was it encoded the contextual summaries in a "language" only it could understand. The reason is how its particular neutral net was structured. So when it decoded the characters I sent it, it would be able to understand. It's freaky that it was able to do it and make it work. Of course there's a lot more to it and I'm on mobile. But I feel that it is an invention it created itself, with me only providing guidance and feedback. In such a case, I could not take credit for the work, only the idea. Should someone try to patent it by truthfully discussing it's origin they would be denied based on that a human did not program / create it.
  • Currently lodging a patent costs thousands of dollars because patent attorneys aren't cheap. But imagine using a generative AI to generate dozens of them - the USPTO isn't equipped for a hundred-fold increase in low-quality patent applications.

  • Just set those AI generators going and publish the results on the web and anything they come up with automatically becomes non-patentable.
  • As AI technology rapidly advances, the potential for generative AI to upend the US patent system is a legitimate concern. With machines able to create complex works and inventions, the question of who owns the rights to these creations becomes murky. It is important for lawmakers to consider the implications of generative AI and to adapt patent laws accordingly. As a student, I have my own view on this question. In the context of essay writing, the rise of generative AI could also impact the need for origi

egrep -n '^[a-z].*\(' $ | sort -t':' +2.0

Working...