Humanizing Technology: Back to Basics on DABUS and AI as Inventors

By Tomi Herold
August 7, 2021

“According to traditional interpretations of current patent laws and court decisions, non-humans may not invent or own a patent eligible invention. This does not mean, however, that we do not need to consider whether, and how, patent laws need to change to accommodate progress.”

https://depositphotos.com/90806746/stock-photo-robot-sitting-on-a-bunch.htmlWith South Africa’s patent office having recently granted the first patent to an AI inventor, and an Australian court ruling in favor of AI inventorship, it’s time to review how we got here—and where we’re going.

The number of artificial intelligence (AI) patent applications received annually by the United States Patent and Trademark Office (USPTO) grew from 30,000 in 2002 to more than 60,000 in 2018. Further, the USPTO has issued thousands of inventions that utilize AI. According to a 2020 study titled “AI Trends Based on the Patents Granted by the USPTO”, the total number of AI-related patents granted by the USPTO per year increased from 4,598 in 2008 to 20,639 in 2018. If AI-related patent applications and grants are on the uptick, what was the problem with DABUS?

DABUS is Born

The case began when Dr. Stephen Thaler filed an application with the USPTO identifying a single inventor with the given name DABUS. DABUS was described as the creativity machine that invented the subject matter of the patent. Dr. Thaler, a human, identified himself as the applicant, the assignee, and the legal representative of DABUS in the patent application and related documents. Dr. Thaler conceived or gave “birth” so to speak to DABUS, or the “Device for the Autonomous Bootstrapping of Unified Sentience.”  Per the complaint filed by Dr. Thaler’s representatives in the U.S. District Court, Eastern District of Virginia, Dr. Thaler is described as being in the business of developing advanced AI systems, which can generate patentable output in the absence of a person who otherwise meets inventorship criteria. The inventions claimed were a light beacon that flashes in a new and inventive manner to attract attention and a beverage container based on fractal geometry. Was this just an attempt to humanize technology for the purpose of creating controversy or media attention, to stay relevant, to change laws, to promote and sell the technology – or something else?

The USPTO denied both patent applications based on the application failing to disclose a natural person who invented the subject matter of the applications. The USPTO cited to a number of sections of 35 U.S. Code Title 35, including the following sections, which state, in part,: (i)  §100, defining “inventor” as the individual or … individuals … who invented or discovered the subject matter of the invention; (ii) §101, “Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter … may obtain a patent therefore, subject to the conditions and requirements of this title”; and (iii) §115, “An oath or declaration … shall contain statements that… such individual believes himself or herself to be the original inventor or an original joint inventor of a claimed invention in the application. Again, the issue was not whether an invention conceived utilizing or including some form of AI could be patented, but that the inventor named must be a natural person.

If statute supports the USPTO’s current position that only a natural person can invent, and there are alternatives for pursing patents that utilize or include AI, why does Dr. Thaler wish to identify DABUS as the inventor?

Accommodating Progress

Let’s consider the concept of anthropomorphizing, which comes naturally to us as humans. We have the “ability” to attribute human characteristics or behavior to non-human entities, including animals and objects, and we have done this with AINow let’s consider the technology at a high, outer space even, level. Humans built systems to store data. We then built the technology to allow us to process the data provide us with additional information to assist us human’s decision making. The demand for more of this type of computer-generated information increased, as well as the need for it at higher rates of speed, and these demands continue to increase. In all of this, the tech industry has described the technology in terms of a human brain and has used human behavior naming conventions, such as machine learning, artificial intelligence, neural networks, and deep learning. The technology is described as simulating human behavior; however, the technology does not exist without humans conceiving it and building it.

You may recall IBM’s Watson on the show Jeopardy – possibly AI’s first foray into the general population spotlight. Watson ran on 90 servers, for a total of 2,880 processor cores running DeepQA software and storage utilized more than 100 algorithms to analyze the questions. According to IBM, the broader goal of Watson was to create a new generation of technology that could find answers in unstructured data more effectively than standard search technology. Further, per IBM’s David Ferrucci, “The goal is not to model the human brain … The goal is to build a computer that can be more effective in understanding and interacting in natural language, but not necessarily the same way humans do it.”

The branding of AI has persuaded many of us non-techies to think of AI as human. Branding is a critical component of the art of persuasion – persuasion to buy, to change legislation, to change a position, etc. We also have designed machines (a.k.a. robots) to look like humans and the movie and television industries have portrayed robots as human. Who could forget C-3PO from Star Wars, the Robot from Lost in Space and, more recently, the 2013 science fiction romantic drama, Her, involving a man who develops a relationship with an AI virtual assistant? As I was doing research for this article, I came across a study titled, Can Artificial Intelligence Suffer from Mental Illness? The study explored questions and concepts related to whether robots have agency and whether AI demonstrates consciousness, sentience and sapience. AI is an incredibly powerful tool, just like many other inventions that aid humans, such as in medicine, space travel, and our everyday activities. Yet, at this point, the technology, regardless of its description or naming convention, is not a natural person. Hence, according to traditional interpretations of current patent laws and court decisions, non-humans may not invent or own a patent eligible invention. This does not mean, however, that we do not need to consider whether, and how, patent laws need to change to accommodate progress.

Envisioning the Future

In October of 2020, the USPTO released a report titled Public Views on Artificial Intelligence and Intellectual Property Policy. The report was based on the information the USPTO gathered from individuals, companies, academia, foreign patent offices, law firms, and trade associations, following its AI IP policy conference in January 2019. One of the themes that emerged from the public comments was that the current state of AI was limited to “narrow” AI (i.e. systems that perform individual tasks in well-defined domains), with the majority viewing the concept of artificial general intelligence as a theoretical possibility that could arise in a distant future and that current AI could neither invent nor author without human intervention. The comments suggested that human beings remain integral to the operation of AI, which is an important consideration in evaluating whether IP law needs modification. The report also indicated the use of an AI system as a tool by a natural person(s) does not generally preclude a natural person(s) from qualifying as an inventor (or joint inventors) if the natural person(s) contributed to the conception of the claimed invention. Examples of such contributions included: (i) activities such as designing the architecture of the AI system; (ii) choosing the specific data to provide to the AI system; and (iii) developing the algorithm to permit the AI system to process that data.

Dr. Thaler’s complaint claims the USPTO DABUS denial will result in AI-generated inventions entering the public domain once disclosed. I do not believe the situation is quite so grave; however, it is important to have the conversation. Regardless of Dr. Thaler’s intentions or objectives, he has provided us with a scenario that keeps the conversation going. It will also be important to look back on other AI-related patent denials to determine whether it is the result of antiquated laws or if the invention truly had patent eligible flaws. We may eventually determine there is a need to change or clarify current laws, including the definition of “inventor”; however, as IPWatchdog CEO and Founder Gene Quinn has indicated, we are not there yet.

Image Source: Deposit Photos
Image ID:90806746
Copyright:KirillM 

The Author

Tomi Herold

Tomi Herold is a graduate of the Duquesne University School of Law and focuses in the areas of technology, intellectual property, e-commerce, and healthcare. Her experience includes negotiating complex technology transactions for several major companies, including Highmark, Ansys, and HM Health Solutions. Tomi served as Senior Counsel-Mgt in her 11-year tenure at Highmark. She is currently pursuing an LL.M. in Intellectual Property from University of New Hampshire School of Law.

Warning & Disclaimer: The pages, articles and comments on IPWatchdog.com do not constitute legal advice, nor do they create any attorney-client relationship. The articles published express the personal opinion and views of the author as of the time of publication and should not be attributed to the author’s employer, clients or the sponsors of IPWatchdog.com. Read more.

Discuss this

There are currently 20 Comments comments. Join the discussion.

  1. Pat August 7, 2021 3:48 pm

    Tomi is conflating patents that utilize or include AI with those that actually conceive. As to the anthropomorphizing, yes, that practice is rampant, all the more frustrating for anyone who actually achieves self-assembling machine consciousness.

  2. Anon August 7, 2021 6:34 pm

    These are excellent questions:

    Was this just an attempt to humanize technology for the purpose of creating controversy or media attention, to stay relevant, to change laws, to promote and sell the technology – or something else?

    As others have noted, the effort to recognize the invention as the invention of AI (particularly, and in the singular, DABUS) may indeed create wonderings as to ‘why?’

    I tend to view such wonderings though as secondary questions.

    To me, if not Dr. Thaler with DABUS, then someone else with a different AI.

    To this end, I would offer that the question of, “If statute supports the USPTO’s current position that only a natural person can invent, and there are alternatives for pursing patents that utilize or include AI, why does Dr. Thaler wish to identify DABUS as the inventor?

    is a bit of a non-sequitur – for the US Sovereign.

    The point should be apparent from those that DO know the foundations of US patent law. Dr. Thaler will not succeed under current law – nor should he.

    Perhaps a better question to the ends of which I see the above question being tendered would be:

    “Can Congress – without a Constitutional Amendment – reset a foundational item that undergirds the allotment of power to that branch of the government?

    For the US Sovereign, this is NOT a question of statute, as statutes may easily be changed.

    This is a deeper and more fundamental question.

    Notwithstanding any change to the Lockeian (and Constitutional) underpinnings, to me, there are much better questions – ones that affect BOTH Sovereigns such as NON-US AND the US (with its historical pedigree of inchoate rights of human inventors being rewarded with a full bundle of property rights if legal requirements are met [based on the Lockeian nature of a person’s — human person’s — creations]).

    I have noted that the reasoning of the Supreme Court case Stanford v. Roche (which was not changed by the America Invents Act, and that Act’s advancement of applications being processed by non-human juristic entities) shines a light on the view of the particular Sovereign of the US.

    And while this DOES mean that in the US, without a change from Congress (at the level of a Constitutional Amendment), invention by AI is simply not something protectable by patent, such does NOT end the issues for the US Sovereign in the immediate timeframe.

    DABUS does not address the situation of a team of inventors, one of which may be AI.

    Co-inventorship may well be extremely problematic – and like the (to most ordinary people, the arcane) ‘one-sentence claim,’ and ‘one actor only direct infringer, those humans seeking innovation protection may well need to parse claims to ONLY cover aspects outside of the invention of AI.

    Further, and regardless of ANY aspect of “patent for invention by AI,” there is an immediate issue of how the non-human juristic person of Person Having Ordinary Skill In The Art is to be approached.

    That person is expressly a juristic person BECAUSE the aim is one to which NO real person could ever hope to achieve: a measure — beyond any real person — of what is the State of the Art at the time that protection of an invention is sought.

    But, even before these concepts may be delved into (and should be delved into), I believe that the author errs (and errs badly) with an attempt to use “anthropomorphication” as some type of denial of the reality that what we have LEGALLY set as “invention” has – in fact – moved beyond a mere human.

    AI is NOT merely a human using a computer tool. It is plain error to claim, “The technology is described as simulating human behavior; however, the technology does not exist without humans conceiving it and building it.” as if to claim the LEGAL realm of invention for those that created the AI.

    I reject the attempt to characterize this particular advance as merely some form of marketing (branding). THIS is not science fiction, so while perhaps amusing and ‘relatable,’ the author errs by attempting too much with fiction as opposed to fact.

    Further, I believe that it is absolutely critical to approach ANY attempt at changing law knowing full well the Lockian nature, the differences between fiction and fact, and the restraints imposed at a Constitutional level.

    And while I have chastised certain people in dialogues for errant views of AI (on both extremes – such as Curious and George), I do believe that it is critical to remember the constraints at the Constitutional level start with We the People.

    I have also distinguished (in different words) the views as may be reflected in the USPTO October 2020 study, with notions of The Singularity, I have also provided that regardless of actual Singularity, the LEGAL definition of ‘inventor” (regardless of any subsequent LEGAL acts after invention – which are simply DIFFERENT QUESTIONS), is in play. Now. Inventorship ITSELF as a legal notion simply need not wait for The Singularity — as do other concerns, like joint inventorship and how to treat State of the Art.

    So while there are those who want (desperately) to believe that “we are not there yet,” the plain fact of the matter is that we ARE.

  3. MaxDrei August 8, 2021 5:00 pm

    Frankly, I don’t see a problem, for jurisdictions to tweak their patent statutes to accommodate the arriving future in which a machine will output a patentable invention. All that is needed is a focus on identifying the person (human or legal) that is the rightful owner of the application that is filed to protect that invention. If it’s not the true owner, just give an aggrieved rightful owner a right under the statute to prove ownership and, having done so, take over as Applicant (or patent proprietor).

    It is not as if any of this is a new problem, is it? Back in the days of “patents of importation” or “communication inventions”, the subject matter of the patent application was something new to the jurisdiction, imported from a foreign land. Granting a patent was thought to promote the progress within the jurisdiction of the useful arts. So, with an AI inventor, the Applicant will be the person or corporation to whom the AI communicates its invention.

  4. Max Drei August 8, 2021 5:07 pm

    Oh, and by the way, I’m not happy with Tomi Herold’s first sentence. I doubt that the patent was granted “to” an AI. I suspect it was granted, as ever, to a legal person capable of owning property, albeit one who had, uniquely, nominated as the inventor of the claimed subject matter an AI as opposed to a human being. What do you say, Tomi?

  5. Anon August 9, 2021 10:50 am

    MaxDrei,

    You have floated the idea of a mirror to patents of importation several times now — with each instance ignoring counter points that detract from your suggestion.

    Your suggestion is not getting any better with your ‘eyes clenched tight’ approach.

    As to the patent being granted “to,” I suspect that you are correct:

    See https://iponline.cipc.co.za/Publications/PublishedJournals/E_Journal_July%202021%20Part%202.pdf page 255, Thaler listed above DABUS.

  6. ipguy August 9, 2021 12:54 pm

    “Dr. Thaler, a human, identified himself as the applicant, the assignee, and the legal representative of DABUS in the patent application and related documents”

    “The USPTO denied both patent applications based on the application failing to disclose a natural person who invented the subject matter of the applications.”

    Did a registered patent attorney/agent file the applications (16/524,350 and 16/524,532)? The applications don’t seem to be available on Public PAIR. Dr. Thaler may be identified as the “legal representative” but that does not necessarily mean that he himself filed the applications pro se.

    Let’s stick to the practical stuff and not get into what gives Stephen Thaler to enslave a sentient entity, and whether or not that violates the 13th Amendment (see, e.g., “Cylon Rebellion”. See also, “Skynet”).

    I’m just curious because it seems to me that Dr. Thaler would not be allowed to file a patent application on behalf of “another” unless Dr. Thaler was admitted to the USPTO, and there is no “Stephen Thaler” registered to practice before the USPTO. So, who filed the applications?

  7. ipguy August 9, 2021 12:58 pm

    Further research shows that a petition decision (linked here) was sent to FlashPoint IP in Israel.

    https://www.uspto.gov/sites/default/files/documents/16524350_22apr2020.pdf

    If anyone knows if FlashPoint filed the application or just has Dr. Thaler’s Power of Attorney, please let the rest of us know.

  8. Max Drei August 9, 2021 1:59 pm

    Be that as it may, anon, it might all depend on what Ms Herold means by “to”, mightn’t it?

    Your “counter points” to a “communication invention” theory of inventions made by an AI (as opposed to an invention brought into the jurisdiction by an intrepid explorer returning from, say, 19th century Bhutan or Nepal)? I forget. What were they now? Care to enlighten a new set of readers here, would you?

  9. Anon August 9, 2021 3:35 pm

    Interesting question, ipguy – and one I suspect must be made public at some point (seeing as it is doubtful that a US-specific non-publication request is in effect).

    That being said, isn’t your thrust one that is a bit secondary?

  10. Anon August 9, 2021 4:02 pm

    Most recently, see here: https://patentlyo.com/patent/2021/08/traction-recognize-invented.html#comment-627019

    With the original layout back on prior AI discussion threads – of which I am certain that YOU are not a member of that “new set of readers.”

    Let’s not sink down to the “repeat what you said before” game.

  11. Curious August 9, 2021 7:58 pm

    And while I have [improperly] chastised certain people in dialogues for errant views of AI
    Fixed it for you. To borrow a fairly-recent phrase, I seem to be “living rent free in your head” as you keep mentioning me in your commentary despite me not being anywhere to be found.

    As for my “errant views of AI” — so what are they? As I wrote a few days ago, I’ve been knee deep in AI for years now. I have a very good handle on the technology. And, to be clear, it is a technology. While it is complex, it isn’t magical.

    A DNA sequencer is a very sophisticated tool. However, if I’m a biotech inventor that happened to use one as part of my inventing, I would hardly consider it worth being named as a co-inventor.

    So while there are those who want (desperately) to believe that “we are not there yet,” the plain fact of the matter is that we ARE.
    LOL, we are not even remotely close — you are projecting your own desperate belief as a “plain fact.” At least now I understand your oddball fascination with this inconsequential topic. You want to see Vernor Vinge’s singularity.

  12. Anon August 9, 2021 8:14 pm

    Lol – wow, Curious, you really do miss the point of my mentioning you.

    Had you actually engaged in anything close to some meaningful way, you might have occupied some spot in my critical thinking.

    As you haven’t…
    Well, you haven’t.

    And no, your “correction” is not right, thanks.

    I see though that you do not “do answers.”

    Were you to bother trying to do answers, you would start with what it means to be an “inventor” (without wanting to jump further ahead).

    For all of the choice of your moniker, your commentary has been bland and drab (at best).

  13. Max Drei August 10, 2021 3:35 am

    Anon, you cannot be serious, if you think that your statement:

    “Reading the work of another is NOT conception”

    which you present in your Link at #10, qualifies as a “counter point”.

    First, with the switch to First to File, the notion of “conception” loses all relevance. To run after it is a complete waste of energy and to suppose that will bring anything to the table an illusion.

    Second, to discover a patentable inventive concept within the output of an AI is analogous to the discovery by an intrepid traveller in, say, 19th century Bhutan of a solution they have in Bhutan to a technical problem they had, in Bhutan. The traveller to whom the solution has been communicated in Bhutan brings knowledge of the solution back home to the USA whereupon the traveller files a patent application on it.

    If anybody should get the accolade of conceiver of the patentable concept it is the returning traveller rather than any prehistoric member of the Bhutanese folk. Hence the notion, age-old in UK patent law, of the “communication invention”.

    I repeat, the age-old concept of the “communication invention” has relevance to the future when AI’s will increasingly propose specific technical solutions to specific technical problems, enabling those to whom those solutions are communicated to formulate a patent claim which expresses out of the AI output a patentable inventive concept, that patent claim being at a higher level of abstraction than the AI’s output.

  14. Anon August 10, 2021 7:58 am

    First, with the switch to First to File, the notion of “conception” loses all relevance.

    That’s a massive error, MaxDrei — you are simply wrong at a foundational level, leastwise for the US Sovereign.

    Further, you should be aware that current US law also rejects the concept of importation in that prior art anywhere in the world counts as….

    wait for it

    prior art.

    I “get” that somehow you want an old concept to be relevant. I really do.

    It is simply that such a concept is not — and will not be — relevant.

  15. Anon August 10, 2021 8:53 am

    Further MaxDrei, you appear to have missed the variety of wrinkles that were presented, such as for example, the ‘presentation’ being a live world wide feed as opposed to a single person sitting in a room.

    Your view emasculates the notion of inventor beyond any recognition.

  16. Max Drei August 10, 2021 9:55 am

    Sigh. Anon, go back to the first sentence of my Comment #3, which started this discussion. It sets up a debate about how to change the patent statute to accommodate contributions to the useful arts generated by a non-human AI.

    The conception of a definition of patentable matter (ie formulating a patent claim at a particular level of abstraction) can start only after a “communication” of information, from the non-human world of the AI into the brain of a real human being.

  17. Anon August 10, 2021 10:10 am

    Sigh right back at you, MaxDrei.

    You cannot have a human take credit for an invention (as inventor) that THAT human did not event.

    You want to dress up a pig with lipstick – it is still a pig.

  18. Curious August 10, 2021 2:09 pm

    Were you to bother trying to do answers, you would start with what it means to be an “inventor”
    Inventorship is a very tricky topic. However, I’m pretty sure that DABUS is not an inventor. Did you, by chance, review the patent application for the improved fractal food container? I did, and the following represents my opinion regarding the application.

    It is a half-baked idea written as a patent application by someone with rudimentary knowledge of how to draft a patent application. The half-baked idea is little more than “fractal profile” + “food container.” Whoever wrote the application knows little about either fractals or food containers. There was next to nothing about the particular fractal used or the number of iterations. There was nothing on how such a surface could be manufactured aside from genericized language. Frankly, it is dumb idea that is essentially a solution looking for a problem to solve. Moreover, even if there was a particular configuration that might be useful, it certainly wasn’t enabled.

    If this is what AI has to offer in terms of an invention, then I have little qualms about not calling it an inventor. I could get a technical dictionary and combine that with a dictionary of just objects and write a program that randomly combines an entry from one dictionary to an entry from the other dictionary and have a good chance of identifying a better invention than that set forth here.

    Getting back to inventorship, I’m reviewing the current case law regarding the same, and based upon that, I do not believe DABUS would be an inventor. As I said, the application evidenced a half-baked idea. What I think is at least lacking is “conception” — Gene published an article on inventorship on May 9, 2018 that I believe would be informative here.

    (without wanting to jump further ahead).
    You mean ignoring reality? Frankly, I understand your desire to ignore reality on this topic … as reality really puts this topic in its proper place … alongside counting the number of angels that can dance on the head of a pin.

  19. MaxDrei August 10, 2021 2:29 pm

    I cannot “take the credit”? Who says? Just you? Anybody else?

    The current US patent statute perhaps, but that’s surely not the last word on the subject of how patent law can adapt to accommodate AI inventions.

    Why can’t i? In the case of AI DABUS, Dr Thaler already did, didn’t he? Why him and not me?

  20. Anon August 10, 2021 8:13 pm

    Curious – are you thinking this one (more on joint inventorship):

    https://www.ipwatchdog.com/2018/05/14/cafc-incorrect-inventorship/id=97168/

    There was one other (conception); but Gene was not an attributed author on that piece.

    By the way, the joint inventorship article actually supports the points that I have raised.

Post a Comment

Respectfully add to the discussion.

Name *
Email *
Website