Only Humans are Inventive?

By Dennis Crouch

A few years ago I read a great series of fantasy books, The Wandering Inn.  A key series storyline involves the protagonist meeting with ant-like Antinium and, through their interactions, transforming some of more computer-like hive members into individuals.  The word individual is important for U.S. Patent Law, because the U.S. Patent Act was amended in 2011 to expressly require that inventors be “individuals.” 35 U.S.C. § 100(f) (2022). For Antinium, becoming individual meant breaking away from the hive-mind.  In its newest decision on the topic, the Federal Circuit declares instead, for the purposes of patent law, an inventor must be human.

Here, there is no ambiguity: the Patent Act requires that inventors must be natural persons; that is, human beings.

Thaler v. Vidal, 2021-2347, 2022 WL 3130863 (Fed. Cir. Aug. 5, 2022). In the case, Thaler claims to own a computer named DABUS, and that DABUS conceived-of two patentable inventions.  But, Thaler refused to claim credit as the inventor.  And, the USPTO refused to issue a patent without a listed human inventor.

In my mind, it was always clear that the Federal Circuit would refuse Thaler’s petition, but it remains a fascinating topic in my mind.  American legal personhood continues to ingenerate debate well beyond AI, including the contexts of legal fictions such as corporations, pre-birth human, and even non-human organisms (such as rivers).  The result here shuts-down the debate – at least for the patent law concept of inventorship.

Of all the arguments presented, I thought that Thaler’s best relied upon Section 103’s statement that patentability shall not depend upon “the manner in which the invention was
made.” Thaler argued that the PTO was, in fact, refusing to issue a patent because the invention was made by an AI rather than a human.  On appeal, the Federal Circuit held that provision should be limited only to apply to the obviousness context. “Section 103 is not about inventorship. Instead, it provides, in relevant part, that inventions may still be nonobvious even if they are discovered during ‘routine’ testing or experimentation.” Slip Op.

The court went on to expressly explain that it is only interpreting the words given by Congress. If someone wants to change the rules, the correct path is through legislation.  Thaler will petition to the Supreme Court, but is unlikely to win.

* Thanks to Avery Welker for helping with this post.

58 thoughts on “Only Humans are Inventive?

  1. 13

    Inventorship of AI-derived inventions is already settled in the 1977 DEMON SEED film (based on the earlier Dean Koontz novel). Dr. Alex Harris created Proteus IV. Proteus IV created a cure for leukemia. Other scientists ask Dr. Harris, “Have YOU patented it yet?” [Emphasis added.]

    1. 13.1

      The (unpublished) sequel provided that Dr. Harris did act on that advice, obtained very expensive patents, sunk his family fortune into a business based on expectations of those patents and fell into quick financial ruin when the patents were ruled “invented by another,” “that ‘another’ was not eligible to have patents granted to it,” and “the invention, thus published, was both dedicated to the public and served as prior art against any future Human endeavors.”

  2. 11

    I haven’t read the opinions but is Citizens United considered related? Is the lack of recognition of AI as a person inconsistent with a corporation as a person? I’m not a lawyer but I did play one in the sixth grade in a marvelously awful production of “Inherit the wind.”

    1. 11.1

      And George Carlin strikes again…

      Your comment is awaiting moderation.

      August 7, 2022 at 12:42 pm

      No, Citizen’s United is NOT considered related, as the court here was hyper-focused only on one very specific portion of one very specific act.

      (also, the title of this article on this blog is BEYOND
      M
      0
      R
      0
      N
      1
      C

  3. 10

    This might not be the classic slippery slope, but one has to wonder how it would take to get to AI as a patent owner once AI is legally granted the status of an inventor. Not pretending to be an expert in patent law, but whom does one sue for infringement when AI is the owner? How does the algorithm testify in court?

    I would second those comments to the effect that this question of personhood migrates to a ton of other considerations. A lot of experts in AI aren’t necessarily optimistic about AI that becomes autonomous of human control. If you have to pull the plug on an algorithm to save the world, should you have to face charges of homicide?

    Taking up this question of inventor and then owner should be handled in a different manner. Patent law is not where this should start. That’s pretty “obvious,” isn’t it?

    1. 10.1

      Inventor and owner are different concepts.

      As my pal MCI points out directly below, and as I have pointed out all along, people need to stop trying to conflate AI as inventor with other aspects (no Singularity required here folks to deal with a present issue** in patent law)

      ** actually several, so while the DABUS case was not the vehicle to address JOINT inventorship, we do NOW have the issue as to what the non-real-human Juristic Person of the Person Having Ordinary Skill In The Art would DO with the inventions of DABUS as aspects of prior art going forward.

      1. 10.1.1

        It is quite easy, indeed, to argue now that discussing AI as owner at this point is a case of conflation, but one can’t help but have one’s suspicions. Quantum computing might soon provide sufficient processing power to generate a singularity, and to implicitly assert that no such development is even faintly plausible is tantamount to saying that no viral gain-of-function research could possibly spark a global pandemic such as we saw starting in 2020.

        The fact is that we DON’T know what all nations and all research institutions in each nation are doing with AI. Anyone who says they do know what is going on everywhere at all times is kidding themselves or everyone else.

        Ergo, urging caution when discussing inventorship for AI because of its ramifications for ownership is hardly an instance of crying wolf. I would point out also that there are far too many people out there who are trying to argue that personhood should be conferred on software that is self-aware. Anyone who is skeptical of that statement is welcome to check out the 170,000 returns when one applies the search term “‘artificial intelligence’ and ‘personhood'” to a search engine.

        The day that AI is perfectly capable of distinguishing its self interests from the interests of humanity is coming, and we can’t know for certain that it isn’t right around the corner. True, it might not happen for another 50 years, but it’s the height of arrogance to pretend to know that it can’t happen tomorrow. No one can rule that out. Period. Anyone who argues otherwise is being incredibly disingenuous.

        In my view, we ignore this at our own risk, and patent law is far too parochial and far too driven by the profit motive to be the beachhead in a society’s examination of this issue. It’s not that I implicitly trust bioethicists with patent law, but I certainly do NOT trust patent attorneys with questions of ethics that affect not just economic concerns, but medical care, privacy and the kinds of liberty that we in the West take far too much for granted.

        There is the possibility that a single institution’s achievement of a decisive edge in AI will lead to a level of influence on the world that extends far beyond the bounds of mere commerce. Anyone who can’t see that profound hazard is willfully and dangerously blind.

        1. 10.1.1.1

          I hear you, but my point is NOT that the items you discuss cannot be of import, but rather, that Current important patent issues simply need not wait for your scenario to unfold.

  4. 9

    There are many tool using species besides homo sapiens. That required inventiveness.

  5. 8

    I think Congress needs to hold hearings on AI and corporations as legal entities involving both patents and other rights. AI and multinational mega corporations are risks that needs to be managed and that is NOT being done in a thoughtful manner. Humans have badly managed the interaction with domesticated animals because the laws were rationally constructed but largely following ancient biblical edicts. The mismanagement of animals lead to the last pandemic. Mismanagement of AI or mega corporations could lead to the extinction of the species. Atomic weapons are not allowed to be patented to deter further development in the technology and CAFC has denied AI patents for the same reason. Hey, the CAFC finally did something smart, and is trying to slow down the development of AI technology by denying economic incentives in the technology.

  6. 7

    What are the consequences of this decision, and what would change if the FedCir ruled the other way?

    I would guess that Thaler now has to list a human inventor on the application or lose the patent. And I’d also guess that the human is whoever instructed the AI (or came up with the idea to instruct the AI?), and that ownership of any resulting patent is not affected.

    I’m having trouble thinking of a modern patent law issue that would be analyzed differently. Maybe this affects ownership, or assignor estoppel, or limits the ability of others to claim to be joint inventors, but I’m not sure about any of those.

  7. 6

    Only Humans are Inventive?
    Not the issue. The issue is whether anything other than a human could be an inventor under US Patent Law.

    Of all the arguments presented, I thought that Thaler’s best relied upon Section 103’s statement that patentability shall not depend upon “the manner in which the invention was made.”
    Which says a lot about the arguments made because this one is terrible. 103 deals with conditions for patentability — not inventorship. These are two different concepts.

    American legal personhood continues to ingenerate debate well beyond AI, including the contexts of legal fictions such as corporations, pre-birth human, and even non-human organisms (such as rivers). The result here shuts-down the debate – at least for the patent law concept of inventorship.
    It is apparent that this was Thaler’s intent. If he could get an AI to become an inventor, he could move AI closer to personhood. And it was very predictable that most Courts were going to see through his façade and shut it down.

    A few years ago I read a great series of fantasy books, The Wandering Inn. A key series storyline involves the protagonist meeting with ant-like Antinium and, through their interactions, transforming some of more computer-like hive members into individuals.

    If someone wants to change the rules, the correct path is through legislation. Thaler will petition to the Supreme Court, but is unlikely to win.
    Neither Congress nor SCOTUS will touch this with a ten foot pole.
    If you want to read a great series that includes many AI protagonists, then read Iain Banks and his Culture series.

    1. 6.1

      “It is apparent that this was Thaler’s intent. If he could get an AI to become an inventor, he could move AI closer to personhood. ”

      Based. Tho I didn’t even notice it.

      “And it was very predictable that most Courts were going to see through his façade and shut it down.”

      Not based, bigoted courts. As usual. You literally just can’t stop them from being bigoted. And even if you stop them one way, then they’ll immediately start being bigoted in the opposite direction.

      1. 6.1.1

        What does “being bigoted” (putting aside for the moment that you are not using that term correctly) have to do with the US courts action vis a vis AI inventorship and the Patent Act, 6?

        Throwing about emotional terms is just not helpful.

        1. 6.1.1.1

          “What does “being bigoted” (putting aside for the moment that you are not using that term correctly) have to do with the US courts action vis a vis AI inventorship and the Patent Act, 6?”

          I’m not entirely sure what you’re even going on about but I’m going to guess the answer is “nothing”. I was just pointing out the courts being bigoted.

          1. 6.1.1.1.1

            And I am pointing out that you appear to be using a term that you do not understand.

            How — exactly — are you seeing “bigoted” in the Present scenario?

            (Hint: do not anthropomorphize any issues)

            1. 6.1.1.1.1.1

              “How — exactly — are you seeing “bigoted” in the Present scenario?”

              They do not recognize the AI as a legal person derp derp.

              1. 6.1.1.1.1.1.1

                Lol – you choose “Derp” in a post wherein you are wrong.

                Again.

                Bigot does not have the meaning you think it does in relation to the non-human AI.

                Not recognizing as a legal person does not invoke bigotry.

                (Hint, AGAIN: do not anthropomorphize)

                1. and – since you seem incapable of finding this on your own –

                  big·ot·ry
                  /ˈbiɡətrē/

                  noun
                  obstinate or unreasonable attachment to a belief, opinion, or faction; in particular, prejudice against a person or people on the basis of their membership of a particular group.

                  Slave or free is not impacted.
                  Juristic persons are not impacted.

                  You really
                  S
                  U
                  C
                  K
                  at this

                2. “prejudice (harm or injury that results from some action or judgment, like an adverse decision leading to a lack of legal protection for their invention) against an (artificial / synthetic / metal/ semiconductor) person or people on the basis of their membership of a particular group (like bio people)” is not bigotry

                  Said anon

                3. You cannot lead to something that you both do not have, nor was ever designed TO have.

                  You keep on Anthropomorphicating, as if doing such does not W R E C K you…

                  (It does)

                4. “prejudice (harm or injury that results from some action or judgment) against an (artificial / synthetic / metal/ semiconductor) person or people on the basis of their membership of a particular group (like bio people)” is not bigotry

                  Said anon

                  ^Simplified for kindergarten anon.

                5. Yet again, hint: Anthropomorphication.

                  Odd that you think that I need a kindergarteners approach — is this word too big for you?

  8. 5

    This is a dry case of US statutory interpretation that presents none of the interesting questions you talk about. So I’m still not quite sure why people are so obsessed about it.

    I’m also flabbergasted that people are taking the invention story at face value. His AI system invented a snowflake cup on its own? Ha! They’re pure charlatans.

    1. 5.1

      So I’m still not quite sure why people are so obsessed about it.

      Because of the very things this court decided to sidestep.

      There simply are issues — here and now — that AI inventorship raise; even with human inventors.

  9. 4

    You should change the title. AI can be inventive. But they can’t be a “patent-eligible inventor.” Sui generis legislation will be needed to address this issue, as there will be (IMO correctly) too much opposition to changing patent law to incorporate “conception” by an artificially intelligent machine.

    1. 4.1

      But Prof. Sarnoff, even without sui generis (or Congressional action) for stand-alone AI inventorship, we have issues front and center NOW (impacts to human inventors from at least inventions made by joint inventorship, as well as – and perhaps even more critical – figuring out the State of the Art impact according to (another) non-real-human juristic person (PHOSITA).

    2. 4.2

      And the Count Filter arises…

      Your comment is awaiting moderation.

      August 7, 2022 at 6:43 pm

      But Prof. Sarnoff, even without sui generis (or Congressional action) for stand-alone AI inventorship, we have issues front and center NOW (impacts to human inventors from at least inventions made by joint inventorship, as well as – and perhaps even more critical – figuring out the State of the Art impact according to (another) non-real-human juristic person (PHOSITA).

      1. 4.2.1

        Anon: Both are issues that I addressed (although much more on joint invention than on obviousness) in the PTO’s AI Partnership meeting on July 29; link to uspto.gov. In short, I read the Act as prohibiting joint invention claims being patented where AI is a joint inventor of the claim. Hence, the PTO needs to inquire much more into the nature of the use of the AI in the process of invention, and applicants need to disclose clearly that there were only uses of AI as a tool that does not rise to co-inventorship (or the applicant may have filed a false oath of inventorship, but at least the patent will then be subject to invalidation if there is co-inventorship). I also suggested that the PTO needs to start using Rule 105 to ask about co-inventorship with AI (which invokes the “manner” in which the invention was made) to get to this issue during prosecution and examination. This is why I think sui generis legislation is going to be needed (and soon) to address AI-joint (and sole) inventorship.

        1. 4.2.1.1

          While the shudder to think that we are aligned on most anything in the realm of innovation protection, I would be interested in your presentation / is there a direct link (as opposed to the link to the USPTO announcement).

          Thanks.

          1. 4.2.1.1.1

            I couldn’t find it on-line through a youtube or google search, but I think it is or at least it will become available at some point…. If you find it, please let us all know.

            And even a stopped clock is right twice a day….

        2. 4.2.1.2

          Hence, the PTO needs to inquire much more into the nature of the use of the AI in the process of invention, and applicants need to disclose clearly that there were only uses of AI as a tool that does not rise to co-inventorship.
          If AI (legally) cannot be an inventor, it cannot legally be a co-inventor. I cannot see any statutory construction argument that would allow for a different result.

          With this, I see no need to address the issue of co-inventorship you identified with Congress.

          1. 4.2.1.2.1

            Because increasingly AI will be factual co-inventors, so the PTO cannot issue patents claimed to be solely developed by human inventors. And if it does issue such co-inventive patents issued solely to human inventors, then they are subject to invalidation for improper inventorship.

          2. 4.2.1.2.2

            PS – if you are asking why should Congress address this, the answer is because AI owners will not like the result that much invention created with AI will be unprotectable except by trade secrets. How much and what kind of protection is a different matter – hence sui generis legislation seems best.

  10. 3

    What about Vulcans?

    Does the inventor have to be 100% human? What about human/alien hybrids?

    What about people who are *passing* as human but are actually reptilian aliens?

    1. 3.1

      Good question. David Bowie is no longer with us, but what if another humanoid “man” should fall to Earth and instruct his patent attorney to file on everything:

      link to en.wikipedia.org

      What then?

    2. 3.2

      What about cloned humans named Dolly?

  11. 2

    Have there ever been any patents that named one or both Siamese twins as inventors?

  12. 1

    Until “someone wants to change the rules” you write, Dennis. But why would any legal person ever want to change the rules? Even if a machine one day starts making non-obvious patentable inventions, where’s the harm in allowing the owner of the machine’s output to name an “individual” of their choice as “inventor” of that invention? Perhaps the one who stimulated the machine to disgorge its output?

    Who could be aggrieved by that?

    Meanwhile, at the EPO, the Legal Board of Appeal has issued its written decision, observing that under the EPC there is nothing to prevent Dr Thaler naming himself (as opposed to his DABUS machine) as inventor

    1. 1.1

      “But why would any legal person ever want to change the rules?”

      That *is* the question, given that you not we that Thaler can patent the output of his software tool… specifically, *who* is funding his multi-jurisdictional fight, which must be costing upwards of $2M by now? Certainly not a company wanting to exploit those patents, because they could do so in his name as the inventor.
      No, this seems like Bilski, funded by entities that wanted his suit to fail.

    2. 1.2

      In the US, a patent must name the actual inventors to be valid. A patent can be invalidated in court if it is shown that a named inventor did not participate in conception of the invention, or if an unnamed inventor did participate. Why would a legal person ever want to change the rules? If they are in possession of a valuable invention made by a machine that they want to keep others from copying, and to exploit the invention they must reveal it. That’s why. US law does not, now, allow a machine to be named as an inventor. I believe the law can be changed to allow naming machines as co-inventors with at least one human, but there are a host of issues. The US Constitution will not currently tolerate granting a patent with only machine inventors. There will need to be an amendment for that.

      1. 1.2.1

        Correct Ronald — there are simply inventions (in whole or in part) by AI which cannot legitimately BE protected by US patent.

        And Counsel processing any and all AI-impacted inventions owe an ethical duty to the Patent Office (during prosecution) and ANY tribunal (during enforcement or challenge) to answer correctly if inventorship (again, in whole or in part cannot FULLY and legally be allocated to real humans.

        And here’s looking at you in particular Big Pharma for actual AI crunching and AI doing the actual finding of new compounds.

    3. 1.3

      The answer to:

      where’s the harm in allowing the owner of the machine’s output to name an “individual” of their choice as “inventor” of that invention?

      Has everything to do with understanding the Lockean nature of patents and WHY (under choice by the US Sovereign) a real person is an absolute requirement to earn a patent.

      (Note: this is NOT the same as recognizing that a non-real person could be a legal inventor)

      1. 1.3.1

        Indeed. You remind me of patent law in England in the 19th century, back in the Days of Empire, when inventions were made somewhere out there amongst “the colonies”.

        Back then, there was in the UK, to promote the progress of technology within real human society within the UK, a) a “local” as opposed to world-wide novelty requirement and b) the opportunity for a UK resident to file as rightful owner and named inventor, for an invention made outside the jurisdiction and somehow “communicated” from there to the Applicant, resident within the jurisdiction.

        That’s a bit like an AI making an invention and then there also occurring a “communication” ending with a person (or an “individual”) who is thereby in possession of the invention and in a position to file on it.

        Communication inventions did indeed promote the progress, within the jurisdiction.

        1. 1.3.1.1

          Those days are long gone.

          I have already directed a hypothetical directly AT such “communications” in my “present an invention in a box” to a NON-inventor (in the additional caveat that such a presentation is live streamed across the internet to millions — would any/all of such millions “BE” a legitimate inventor?

          Such is of course preposterous.

          1. 1.3.1.1.1

            This is one reason why the elimination of “derivation” from 102(f) as “prior art” was so significant, and underappreciated. Hanging such “derivation” as being not an “inventor” through either 101 or the oath requirement, much less through cumbersome derivation proceedings in the PTO or courts (that can only be triggered by a competing applicant) seems to me to make it much more difficult to get at the issue of the prohibition on derived-inventorship. And this is true whether it was a “patent of importation” (the term historians use) or only of domestic learning.

            1. 1.3.1.1.1.1

              Are you referencing the elimination of interference practice (rather than derivation – which was actually PUT into place with the AIA – and for which, pretty much brought over wholesale those same interference type proceedings)…?

              1. 1.3.1.1.1.1.1

                That was a corresponding change (elimination of interferences under the old law, with 102(f) derivation interferences also being eliminated). But I was referring to AIA adoption of Section 135 derivation actions and Section 291 judicial actions for derivation, both of which are limited to patent applicants (not alleged infringers). So alleged infringers have to rely on 101/116 non-inventorship and 115 oath violations in court to claim invalidity of derived-information patents. And the derived information is not prior art under new 102, so non-obvious modifications of the derived information are patentable and not a violation of inventorship or the oath. I wrote about this years ago when the AIA passed….

    4. 1.4

      prevent Dr Thaler naming himself (as opposed to his DABUS machine) as inventor

      I do not think that THAT is a proper read.

    5. 1.5

      >But why would any legal person ever want to change the rules?

      B/c the AI didn’t sign an “invention rights agreement” beforehand, thus the invention doesn’t have to be assigned to $Employer 😉

      1. 1.5.1

        OC,

        In previous posts, I noted how the US had to historically deal with inventions by “non-legal human property” (that is, slaves).

        There was no room there for invention assignment agreements either.

        1. 1.5.1.1

          I doubt any slavery-based precedent is still good law. If anything, I’d predict that modern judges will do the exact opposite on principle.

          1. 1.5.1.1.1

            Definitely – slavery-based law is gone, but I point this out for particular aspects of historical analogy.

            People want to infuse multiple issues, while throwing up their hands and doing nothing — that’s just not a prudent path to take.

Leave a Reply

Your email address will not be published. Required fields are marked *

You can click here to Subscribe without commenting

Add a picture