Patent Act requires an inventor to be a natural person, not an AI [pdf]
cafc.uscourts.gov
cafc.uscourts.gov
If AI can create patents, can AI also publish prior art? It seems only fair, right?
But, in that case, the same AI technology used to file 1,000 patents would be used to create a million trillion publications of prior art. If an AI can produce something worthy of patent, then a freely published version of the same would surely count as prior art. We'd end up with the "great tome of prior art" a 500 terabyte download filled with patent worthy ideas but freely published to establish prior art. This is just one of the many things we might find in Pandora's Box of AI patents.
[1] https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2399580
I have always struggled to express exactly what my issue is with certain software patents.
With this paper I now have the vocabulary (and examples) to explain what I mean to someone outside the industry.
The lawyers don't care, the PTO doesn't care, and industry executives don't care. The only answer is to get a bill in Congress, and start asking every candidate for Senator or Representative if they support it.
And make it clear that your support for them depends on a Yes answer.
Like you, I think preserving the status quo until those questions are answered is the right thing to do.
Turtles all the way down.
What would be the difference?
Maybe it’s not the AI we should get rid of but patents?
Given the AI's output using the trained weights of its models were in part deduced using their works as training examples, one could argue that creators of the example pictures should have some portion of rights on any output from the AI Model... maybe...
As evaluated by a patent office with a profit motive to find it non-obvious. No, I'm not joking. USPTO is funded entirely from fees.
Of course, once you've exhausted your two rounds of rejection with them, you have to pay again for another two, or appeal it, so they make money arguing with you, too. But yeah, probably more $$$ to issue you some worthless government paper.
What if it can only be produced by the tool through a highly nontrivial prompt? Prompt engineering is already a field, and there are already people selling prompts.
Maybe models like dall-e are going to be considered a new kind of very high level programming language. Just because you can obtain a certain result with a C compiler (given some specific "prompt" - that is, the C source code) it doesn't mean the result isn't novel.
With the proviso that things may have changed in the last 6 years when I haven't been working in patents:
Your experience is not representative. While in Patent Litigation at Google, I looked at literally hundreds of prosecution histories, and an obviousness rejection figured in almost 100% of them. That's much more common that an anticipation rejection, i.e. ONE reference.
And any lawyer will tell you that if they allow your application on the first try, you didn't ask for enough.
If you tell us the numbers of the patents you worked on, I can look up the histories. Or you can, if you know that one weird trick :)
(Actually, it's several weird tricks.)
The PTO rejects; the lawyers argue & amend; the PTO rejects again making it Final. But it's not Final! You can pay for another go-around.
What you think of as a "rejection" is really Abandonment. They keep rejecting, and finally the lawyers give up. They hate doing that, because they're getting paid to get you something.
What kind of argument is that? That's not even something anybody is debating about?
Prior art just nukes the novelty of a claim in a subsequent patent written by anyone, including the whole patentability.
The source of the prior art is not important.
Related, “Using Artificial Intelligence to Mass Produce Prior Art in Patent Law” (2021):
https://vanderbiltlawreview.org/lawreview/2021/03/the-librar...
Indeed if the AI isn't the author it's just a tool, and whoever is using the tool should be the author!
Even if computers were just as intelligent and sentient as humans they have no rights or recognition by law. They are not members of society, citizens or entities that have rights and obligations under rule of law. A document/proceeding that uses a non-human entity (I would exclude animals but even they have some rights) is by default invalid.
But, I say what should be under reason and logic which are not important to many judges or this society as a whole as evidenced by asset forfeiture and how they consider property a subject of prosecution as if property has any obligation or capacity to follow the rule of law. Essentially, the law is whatever arbitrary thing judges feel like with no rhyme or reason other than "this won't get me disrobed/jailed/disbarred". Companies are people, propery is people, people are property if they commit a crime (but not the kind of property that has rights lol). Bees are a type of fish in california (really.), you get the picture the assumption that the law is reasonable is false even though the "reasonable person" doctrine is what supposedly keeps the US legal system running. So in reality, this ruling just reflects what the judge felt like, appeal to a higher court and they might decide AI has more rights than minorities.
[1] https://news.bloomberglaw.com/us-law-week/only-humans-not-ai...
Corporations are a group of people working together for a common goal. “AI”s are made up of zero humans.
It makes sense that a corporation can own something, because that is just a legal proxy for humans owning it. It doesn’t make sense for an algorithm to own anything.
If it’s “any of it is human”, how about AI cooperating with a human? Or perhaps connected to a human brain?
But does it have to be a brain? What if it’s, say, a human appendix? It would fulfill the same role it does in humans.
EDIT: Some people apparently don’t realize that to corporations, humans are just another asset, like buildings or software. And they get replaced much more often than buildings and some machines.
Corporations aren’t humans - they use humans. You could argue it’s usually humans who makes decisions - however, 1. Not always and you can’t tell from outside; 2. Even when it’s humans, they are based on data prepared by a machine; and 3. Even law handles consequences of those decisions differently from decisions made by humans for themselves.
The legal definition of a person is very clear. Even the logical definition is actually quite clear: anything that can possibly exercise most of the rights of a human in a meaningful way can be considered a person.
Can a horse own a piece of land in the same sense that a human does? Not really, as a horse simply can't exercise the right to sue someone who would infringe on its property rights even if it were granted one. Sure, a small child or an adult in a coma may be in a similar situation, but the child may eventually grow up and become able to exercise such a right, and the adult may eventually wake up from that coma.
A machine of any kind that actually exists today is far farther from being able to recognize and exercise any human-like rights than the horse is, in fact. So it's even more absurd to assign copyright to a machine than to an animal.
Tell that to Blake Lemoine. /s
Indeed: “The reason for the term "legal person" is that some legal persons are not people”
(A collective entity is a technical term that does carry the implication of being composed of humans)
Connecting a human to a machine doesn’t change the rights of the human, and it doesn’t give any rights to the machine.
These are all assets of the corporation; they are not the corporation.
Figuratively, maybe. Legally, no. This conversation is about specific legal terms, so figurative usages do not apply.
The only way they can be equated is if you make generalizations so broad and metaphorical as to be legally meaningless.
> It doesn’t make sense for an algorithm to own anything.
Current AI might be that. I don't know when AI will end up with a rich inner world of experiences, but I bet it won't be the exact same moment the law recognises it as such.
I'd go even further: even if it somehow becomes possible to prove absolutely that some AI is sentient (which would be a remarkable development in the philosophy of mind), I'm fairly sure people will argue against such a proposition at least a decade after an AI passes such a test.
Because you ended that with a presumption "no", regardless of if I said "yes" ("ja") or "no" ("nein") you might well take that as an agreement, but "doch" is (when used as a particle) always a disagreement with the main claim: https://en.wiktionary.org/wiki/doch#German
So:
Doch, a perfect copy of a human mind but in software would be a counter example where the human intelligence itself would imply to me that it has personhood (others will argue about p-zombies, and rich-inner-world consciousness is not well enough defined to convince people in general even with this, which is where I was going with my original point).
But it wouldn't matter if the answer was "yes", because I was very careful to write about sentience and inner world experience rather than intelligence with regard to rights.
If you wish to argue that intelligence is necessary for creativity, then existing AI must count as intelligent, but (A ⇒ B) ⇏ (B ⇒ A) so don't mistake that for a claim for the reverse.
Natural rights are partially rooted in some religious tangents that I don’t really wish to go down either, but I don’t know of any software other than TempleOS that is claimed to have come from god. For pretty much all AI algorithms, I think people agree they are man made.
For me the issue with recognizing that AIs can have copyright is the difference in how they are bestowed. IANAL, but you actually have copyright on an expression of something the moment it is expressed, even without any slow legal paperwork, whereas patents only mean something once the patent is officially reviewed granted, no? So the unbounded rate of AI's expressing all plausible content (a thousand monkeys but much smarter) and thus preemptively stopping humans from getting copyright on it, seems especially awful.
Imegine the courts were to recognize some algorithm as the legal owner of a copyrighted work. What exactly would that mean?
If someone were to reproduce that work, how would they seek approval from the algorithm? If they don't, how would the algorithm sue them? If it did and won, how would they pay restorations?
Would any copy of the algorithm hold the copyright of that work, or only a particular one? If all copies do, and you infringed their copyright and had to pay damages, how much damages would each copy of the algorithm receive?
The concept of an algorithm holding copyright or a patent or nay other kind of right over something is in fact logically incoherent if you examine it even slightly.
Now, if in the future there would be some form of advanced AGI that does have the ability to understand and use such rights, the situation may change. But nothing that exists today comes anywhere close.
https://en.wikipedia.org/wiki/Monkey_selfie_copyright_disput...
The more interesting question I think is whether any of the creators of the training data used by the AI have any claim over the generated artwork.
(at least in regards to black box models)
The difference is that Photoshop doesn’t create 2D works any more than Blender creates 3D works. You may be ascribing too much magic to Photoshop.
Feels like we covered this back when we decided whether sampling music creates a derivative work.
However, in the case of things like industrial patents or drug patents, where there is considerable cost in shepherding a design through a complicated regulatory framework, it makes sense to give the company that fronted those costs some kind of temporary protection from unfair competition. If you look at patents under that lens, it makes economic sense that an AI generating an algorithm over the course of hundreds of hours would be patent-worthy.
Like you mention, there are however proponents, who feel progress can be prohibited by regulation. Chemistry patents are notoriously nuanced.
Anyway... We're essentially spelling out the [1] Big Tech vs Big Pharma arguments.
[1] https://www.ft.com/content/6c5b2cca-ae8b-11e7-beba-5521c713a...
One of the reasons why IP came to be was to incentivize people to publish details of their inventions, instead of trying to keep them secret.
Not everyone did (famously, WD-40 formula is still a trade secret), but most innovators did, and other people could inspire themselves from such published works. Sometimes in areas that didn't infringe on the original patents.
A public knowledge whose use is artificially restricted by law probably still beats a culture of secrecy.
That's a design patent, which is just a misleading name for something close to a trademark.
No patents are there for the inventor to go through the process of trial and error for indefinite periods of his life in the hopes he will at one point find the iconic invention. It's about there being a bright light at the end of the tunnel that can be seen all the way from the entrance.
if an idea is protected by such laws - you are limiting thought.
I do not support such artificial thought limits.
In addition - arguments for it allowing innovation or promoting innovation...
no: innovation comes from innovation's own sake - patens and copywrite only incentivize spending resources from other's thinking the wrong thoughts... patent trolling... or milking patents for what they are worth...
Humanity would be a lot better off with a copyleft view of intellectual property. An idea in another person's head does not diminish your idea in your head... people need to stop thinking that thought itself is a zero-sum game.
I upvoted you, but I also have a counterpoint. If you do not protect the value of thought, then the differentiator among competing businesses boils down to cheap assets, relationships, and labor. Because everyone is sharing the same designs and processes the company who can out-cheap the competition wins. Innovation is not valued because it's cheaper to emulate what works.
If, 20 years ago, you were to write a program that used an optimization algorithm to design the optimal cup for drinking coffee, the patent would go to you, not the code you wrote. Today, you could do the same thing but call the program an "AI" and perhaps get a different outcome.
I can still have my GPT4 minions writing 100 patents per hour and corresponding with USPTO on my behalf, then as a natural person, collect the proceeds netted from my robo-patent-trolls.
Or is the USPTO concerned about AGI doing the same and robbing us fleshlings of our I’ll-gotten gains?
Ideology and precedent. And proximate to both, personal vanity: someone responsible for an AI that was politically legitimized in a role usual reserved for "natural persons" will likely have made some lasting mark in history. Some people are very motivated by that kind of thing.
If I were the one arguing for sake of vanity, I’d go after the legal name policy vs AI. Maybe that was considered or trialed before.
AI can generate text, it can also review and conduct tasks such as prior art search or support analytics relative to exploitation discovery.
Less if a cat-and-mouse if patent troll investments falter when their mass-generated suite of applications are all rejected.
Experts: AI should be recognized as inventors in patent law - https://news.ycombinator.com/item?id=31542285 - May 2022 (3 comments)
AI cannot be the inventor of a patent, appeals court rules - https://news.ycombinator.com/item?id=28640111 - Sept 2021 (239 comments)
Only Humans, Not AI Machines, Can Get a U.S. Patent, Judge Rules - https://news.ycombinator.com/item?id=28405333 - Sept 2021 (7 comments)
South Africa issues world’s first patent listing AI as inventor - https://news.ycombinator.com/item?id=27995313 - July 2021 (75 comments)
EPO and UKIPO Refuse AI-Invented Patent Applications - https://news.ycombinator.com/item?id=21990346 - Jan 2020 (39 comments)
The court simply ruled that "the Patent Act defines 'inventor' as limited to natural persons" and it's up to the legislature, not the court, to decide if that should change.
I'm not sure what that would exactly look like, but one of the things I like about Monte Carlo is that it works much more like how humans actually behave, and in theory when pressed on 'why', it could also behave more like humans behave: When challenged you (or at least I) run a more detailed simulation to pick some particularly juicy counter-examples. And if anywhere in there I sense self-doubt (why did I make that choice? Is it sound?) that may trigger a more exhaustive simulation. Turns out my pupil was right/wrong and now I understand the problem better myself.
What bothers me with respect to AI and patents is that if you did this, then you basically have a human rubber stamping an AI design. Eventually you'll end up with more subtle systems that resemble police work: this evidence is inadmissible in court but it does eliminate a set of suspects, so now I need to build a chain of clean evidence that points to an arrest. The latter might involve some skill and sophistication on the part of the human. The former just needs someone who can pass a sniff test on materials science or mechanical engineering who's willing to lie for money.
I see this Patent Act as a signal that USPTO is primarily concerned with a quality patent system; as opposed to a loose philosophies that would degrade patent quality but likely heighten patent application submissions, and subsequent flood of trials that would all benefit their revenue streams.
I am on 50+ US patents, and a practitioner of artificial intelligence for about 40 years. I see AIs becoming partners in writing patent applications. Corporations that own AI resources should really be able to use them for legal and ethical things.
Off topic, but it is ethics that introduces a wrinkle: in an ideal world widely accepted ethical norms would be codified into law - not much chance of that happening given the powers of special interests.
[1] https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2399580
Another important detail is also is being skipped here. What is AI? Current "AI" is literally basic algorithms with bunch of if statements. If my software spits out something that I can use, is that considered an AI?
A person who points the tech at the right problem and configures it to get an outcome is the rights holder.
For your question, I think you will be within your rights to list yourself as the inventor of AI-generated invention, provided you were the first to do so and file an application that was accepted by the USPTO and defend the claim from anyone else who says they were also able to generate a similar work with a similar or same prompt.
It is an unsettled question whether a patent would be granted upon inspection by the USPTO.
IANAL and this is my common sense reading of the ruling.
The inventor can not be an AI. You could claim you invented something using the AI as a tool; I don't think that has been tested properly yet -- it would seem to depend on how much input you have. For example if you asked an AI to design a chair, and the AI comes up with some new form of wood joint that serves a technical purpose, then you don't really have the right to hold a patent on it ... it's likely no one would know if you didn't tell them.
Likewise, does a hammer build a house, or its wielder?
To quote the supreme court from Pembina Consolidated Silver Mining Co. v. Pennsylvania (1888):
> "Under the designation of 'person' there is no doubt that a private corporation is included [in the Fourteenth Amendment]. Such corporations are merely associations of individuals united for a special purpose and permitted to do business under a particular name and have a succession of members without dissolution."
Clearly an associations of individuals should have more or less the same rights they have individually when they come together. The right to association is at the beginning of the bill of rights.
If that's the case, then why does the formation of the corporation occur at all? It's not about "easier bookkeeping", it's because the corporation's existence shields its owners from individual liability. That's why it exists, rather than just non-corporate agglomerations of individuals. The corporation itself is a new legal entity, whose existence changes the legal culpability of the owners, and because of that, it makes perfect sense to me that while the owners would not lose their individual rights by forming a corporation, neither do they cede or grant those individual rights to the corporation.
> The corporation itself is a new legal entity
Yes clearly.
> neither do they cede or grant those individual rights to the corporation
I'm sure you would agree that corporations can enter into contracts. This is an individual right granted to corporations. They also have due process rights, unless you think governments should be able to arbitrarily intervene in their affairs. Due process is an individual right granted to corporations. Corporations can sue or be sued in courts of law, again this is an individual right/responsibility. These are all practical things. It would be unwieldy for owners to all have to be individually party to any contract or suit, especially in cases where ownership is distributed through stock. The legal fiction arises out of practicality, centuries of history, and collective exercise of individual rights(as owners of a corporation we all own X, or enter into contract with Y).
Legal personhood means that a corporation can sue, be sued, and enter into contracts. That is all.
That’s really just an issue with the 1st amendment that should be further amended, not an issue with the concept of corporate personhood. The concept of “natural personhood” exists to handle cases where the differentiation is necessary, and it’s the legislature’s job to use it.
Nobody in that group of people has lost anything. If 10 people walk into a room, you have 10 legal persons with 1st Amendment rights in that room. If they organize into a corporation, you have 11 legal persons with 1st Amendment rights in that room!? Without Citizen's United, you simply still have 10 legal persons with 1st Amendment rights. Who's losing anything?
Corporations don't have first amendment rights, but the Individuals do.
The idea is that just because you work with someone else to put up posters, your right to put them up doesn't go away.
Same if you hire someone to put up your posters.
That's all that corporate personhood means. People get very hung up on the language.
People get very caught up on the legalese but it is perfectly logical when you move past that.
Yes, if 10 people form a corporation, that is 11 entities recognizable by a court system that has first amendment rights. All 10 humans can criticize the president at home, and you can do so work in an official capacity without your corporation facing consequences.
It absolutely is not. It creates an entity to which actions can be attributed but the responsibility for its actions does not and generally cannot extend to "people who make up the group".
Somce of us believe that even though there are benefits to society from allowing such entities to exist, these entities should not be accorded the rights of personhood.
[ EDIT: should NOT be accorded ]
If a law shouldn’t apply to human person but not a corporation, the law simply need to specify “natural person”. Corporate personhood doesn’t interfere with this.
> which does not make the distinction you refer to
Maybe it should, then. Congress can change it.
"the law", yes. But the US constitution is not "a law".
> Congress can change it.
Proposing changes to the US constitution as a method of dealing with contemporary policy and procedural issues deserves a name ala Godwin's Law. I mean, it's not that I disagree with you, it just that it's not going to happen.
In that argument each individual people keep their 1st amendment right, but the group just doesn’t get a new one.
Put it another way: suppose there’s 3 physical people, why should we allow them to combine into a maximum of 8 legal people ?
The government can’t fine Alice for political speech.
The government can’t fine Bob for political speech.
AND the government can’t fine “Alice and Bob, Inc.” for political speech.
> Put it another way: suppose there’s 3 physical people, why should we allow them to combine into a maximum of 8 legal people ?
Because they can belong to more than one group. I own parts of dozens of corporations along with as many as hundreds of thousands of other people. When one of them does something wrong, they can be sued with one name, rather than suing me and each of their thousands of other owners individually for their portion of the investment.
This is where opinions differ. Because I am absolutely fine with the concept that "Alice and Bob, Inc" has none of the rights of personhood (really, citizenship), and that if there was a statute that allowed for prosecution of political speech, it could be used to silence "Alice and Bob, Inc." even though it cannot be used to silence Alice or Bob.
Notice that as a matter of nuance, I differentiate between for-profit corporations and other forms of corporations. I would accord some (perhaps all) of the speech rights of persons to the latter, but none of them to the former. By incorporating with a structure created to facilitate the generation and distribution of profit, you acknowledge that the corporation thus created does not have the rights of personhood.
Or, to seize their property without a warrant. Could you imagine if the Trump administration had the legal right to seize property at Twitter, WaPo, etc?
In the case of Citizens, the law involved covered who can spend what on election related advertising during certain periods close to an election. The law was acknowledged by everyone to not impact the individual right to free speech, but in the lower courts it had been ruled constitutional to inhibit corporations of various kinds in this way. SCOTUS said "nuh-uh", and ruled that corporations have the same speech rights as natural persons and that their right (in this case, to publish a book about Hilary Clinton) could not be restricted by such a law.
I don't actually disagree entirely with Citizens at all, I just wish that SCOTUS had limited the ruling to non-profit corporations.
"That decision and subsequent lower court decisions have led to SuperPACs, which allow corporations, unions and individuals to make unlimited contributions, pool them together, and use the money for political campaigns."
https://www.npr.org/2012/02/23/147294511/understanding-the-i...
This goes well beyond what the law was intending to limit for influence on companies on politics. And that was the point of the whole supreme court case.
I for one am not happy with unlimited money spent by huge companies who are using shareholder money to fund political contributions that further their own motives. But hey sure, let's argue that AI systems shouldn't get patents, but corporations should be free to influence politics as much as they want with unlimited money.
It's not "silly" at all. Corporations can do things that people cannot - like cause damage, death, destruction and not involve their owners in the responsibility. It makes sense that the owners can do things that the corporation cannot, like have free speech rights.
Besides, nobody was suggesting that the individual owners lost their individual speech rights by incorporating, merely that incorporation creates an entity that ought not to have free speech rights.
Unless you're going to ban or greatly limit political contributions across the board by deciding they aren't really speech, it's not clear drawing a line between natural people donating and corporations donating makes a lot of sense so long as both have freedom of speech.
What was ruled unconstitutional were laws that said you can't spend money on an election if you do it via some corporate structure.
If a newspaper prints an article, is the government allowed to say it can't publish it because the newspaper is a corporation? Obviously not. Is the corporation now a person with free exercise? No, the journalists who wrote the article are the ones with the rights.
If they don't want any change in their legal rights and responsibilities, it is simple - don't form a corporation. If they do, I'm fine with denoting the newly formed entity as something other than a person, and with not giving it the same rights as its owners.
Practicality of the legal system does, given modern finance. Any Fortune 500 company has hundreds of thousands if not millions of owners. Listing them all on a contract would be impractical and suing them would be equally impractical.
> I'm fine with denoting the newly formed entity as something other than a person, and with not giving it the same rights as its owners.
Well, we already do that. Corporate persons do not count as natural persons under the law. If you’re frustrated that a law doesn’t specify that, that’s a problem with that law not a problem with recognizing corporations as named legal entities.
More fundamentally, the first amendment is not a law, and the Constitution in general does not make the distinction that you are referring to. I am fairly certain that had the law in question for Citizens made the distinction you are referring to, the legal challenge would still have made it to the SCOTUS, and they would still have ruled the way they did.
And yes, the Bill of Rights is law, and as such it is within the legislative's power to change.
Even if an AI were to become a person at some point in the far future, they would still be an artificial intelligence/person, by definition: they are man-made, not naturally born. That wouldn't make them lesser, just different.
except, i don't know if they noticed, but we _keep asking these questions_ daily. the time isn't in the far future and people seriously need to start thinking about this. humans are so stuck in their linear perception of time, we forget that the AI we take for granted now are only months old, and two years ago their abilities were unheard of.
When we'll have anything resembling an AI that actually wants things, it will make sense to discuss about its rights. The current state of the art in AI is that we can generate sentences that match a given context. There are a few billion steps left from here to there.
The fact that LMs can seem like they generate meaningful sentences is nothing more than a nice trick. My bet is that they will turn out to be a dead-end on the path to actual language processing (that is, the ability to read a book and use the information therein to achieve a goal).
That’s a big “if” if I’ve ever seen one. What are your criteria for sentience? Or life, for that matter?
With laws like these we would be incentivizing sentient silicon life to hide itself. Can we not be adversarial by default please?
The question of what legally counts as sentience will be determined piecemeal and politically as the issue becomes relevant, and revisited as new political and ideological regimes take hold and as new innovations challenge prior determinations.
All along, there will be people who are convinced that sentience only applies to humans or biological things or whatever, and people who are convinced that it applies to anything that can perform a certain way under certain conditions, and then also people with all sorts of other convictions.
The issue will not be settled in your lifetime, so if you're so invested in your belief that this already feels like discrimination to you, plan for a lot more frustration and tumult in your future.