Preparing for Automated Examination

Associates around the country today are drafting motions, patent applications, and other documents using some version of ChatGPT.   Of course, If I were a judge or examiner, I might also be interested in using AI to help facilitate my decision-making.  ChatGPT is good for that as well and can provide a reasoned structure, including identifying of prior art and obviousness standards.

If you recall your last writing course, the professor repeatedly focused attention on the audience. “Who is the audience you are looking to persuade?”  Moving forward, it appears that we really do have a new audience for our writing — AI advisors.  This calls to mind the great tape recorder scene from Real Genius (1985). Members of the class began just leaving tape-recorders to record the  boring lecture; eventually, the professor also just brought in his own taped lecture to speak to the machines.

Automated Examination: On the patent prosecution side, we may shortly be in a situation where AI systems will be able to conduct patent examination at the same level of quality as a typical human examiner.  It seems obvious that we should rely upon the AI to at least facilitate examination.  The bigger question is whether to eliminate the first-level human bureaucracy all together and move to an automated examination system.  The benefit of that process is that it could be done very rapidly (immediate allowance / rejection) and potentially at a much lower cost.  Certainly, PCT systems could move toward substantially lowering costs with initial search reports and examination being done automatically.

What do you think? 

112 thoughts on “Preparing for Automated Examination

  1. 24

    Rather than start with the hardest possible application, what about a simple automated prior art search using an AI search engine at the time of filing?

    Patent filers would have the option to run a search using an automated tool and explain why the two closest references from the top ten presented by the tool are not prior art. The USPTO could encourage this by providing preferential queues for filers who choose this option.

    Benefits:
    – Encourages patent filers to perform a prior art check before filing, reducing the filing queue and improving the quality of patent filings.
    – Help independent inventors from wasting money on inventions that are not novel and attorneys who might encourage worthless filings.
    – Demonstrate how AI might be useful in the patent process to the public and patent examiners
    – Positioned as an option, it does not force anyone to use the feature

    PQAI is an open-source AI prior art tool that could provide this function. http://www.projectpq.ai

    1. 24.1

      Your terms and conditions do NOT carry enough safety for practitioners.

      This is an open invitation for malpractice.

      1. 24.1.1

        The USPTO could implement their server for this proposal, so this would not be an issue.

        Regarding our public site, we would be open to any suggestions you might have to improve the terms and conditions.

        1. 24.1.1.1

          Does not resolve the issue — a client’s input is absorbed into your process and shared with others.

          That this may happen through the USPTO server does not stop that “sharing” beyond the immediate client.

          This is a systemic problem that is inherent in what you offer.

          1. 24.1.1.1.1

            You are assuming that the AI software is learning from each user interaction. PQAI does not use queries from users for training. We trained PQAI using a separate dataset. No information provided by the users is shared either directly or indirectly. Therefore the “systemic problem that is inherent in what you offer” does not exist here.

            Returning to the original idea of using AI to create an initial automated search to ensure all filers do their homework before filing could significantly improve the quality of filings and benefit independent inventors.

            1. 24.1.1.1.1.1

              Excellent – yes, I did presume as you identify.

              If indeed there is no data leakage, please ratchet back the alarm of my post.

              This DOES sound then more like using a trusted third party source – do you sign NDA’s?

  2. 23

    Reportedly the CAFC granted another writ of mandamus and again ordered the WDTX Waco district court to decide a pending motion for intra-district transfer (to Auston TX) before proceeding to substantive pre-trial issues including Markmans. In re Apple Inc., Case No. 23-120 (Fed. Cir. Mar. 6, 2023) (per curiam) (nonprecedential).
    [One could argue that this is humorously relevant to the topic here of “artificial intelligence,” since the Fed. Cir. had already strongly made that same legal point in prior mandamus decisions against the same court.]

    1. 23.1

      … any update on the rather tongue in cheek reply by that aforementioned Western District Judge poking fun at the CAFC “yep, you are right in your rewriting of our district rules” case?

  3. 22

    What logic will these fancy new algorithms use to determine what I’m trying to say? Will they alert me in some meaningful way before “correcting” the words I’m typing in instances where the proposed correction radically changes the meaning of what any person with a loose grasp on the English language would actually want to communicate?

    Answer: nope.

    It’s a con. And anybody who tells you otherwise is eyebrow deep in the b.s.

    1. 22.1

      Perhaps you forgot: “And keep off my lawn.”

      Perhaps.

      Perhaps not (there is a strong likelihood that you aren’t aware of even having a lawn from your vantage point of your mother’s basement).

  4. 21

    I sure hope we can get auto-examination. Though I doubt it’s going to be all that great in the next 10 years or so.

    In any event, bros, I was just reading this wiki article on this Stilwell character from WW2 china/burma theatre. What a story. He’s literally perhaps the beginning of the reason why washington stopped supporting the chinese nationalist gov. vs. the commies back in the day. Apparently he had his reasons for doing what all he did. Nationalist gov. at that time was super corrupt according to him as well. Quite a tale though.

    link to en.wikipedia.org

    1. 21.2

      “washington stopped supporting the chinese nationalist gov. vs. the commies back in the day”

      Dennis Crouch’s patent blog, folks.

      1. 21.2.2

        When D’s “patent blog” is infected with commieism as much as the next dept/outlet/etc. it all ties in. New spe’s just dropped. 5 WOC, 4 WW, (yes 9 women), 4 WM, 0 MOC. Bit odd on the numbers there isn’t it? I’m sure there’s no bias!

        In any event, the latest in leftist advice dropped.

        link to msn.com

  5. 19

    I’ll say it again: a computer is a computer and this term “AI” is a salesperson’s gimmick in exactly the same way as the term “self driving” car got a lot of gullible people to stroke the diseased ego of Empty Husk.

      1. 19.1.1

        He has a fake Patent, and I have70 years of being robbed. I do all the un uncovering, and in the mean time you ckaim I am stupi…d, mentally challenged, and with a low IQ. NOW WHAT HAPPENS?
        Do I get 70 years and what was taken from me?

    1. 19.2

      There is some truth to this. AI is just information processing for tasks that humans consider to be indicators of intelligence.

      At the end of the day, AI is just information processing.

      But the rub is that our brains are nothing more than information processors.

    2. 19.3

      I’ll say it again: a computer is a computer and…

      And I will invite you again to engage the Grand Hall experiment with a modicum of inte11ectual honesty.

      Funny that — you never have.

  6. 18

    Just keeping up with last year of absurdity regarding this silly trend in computing: we can laughably contemplate a future where a computer drafts and files a patent application for improving a computer’s ability to examine (or prosecute) a patent application, whereupon that application will be examined by another computer. Similar story with “improved” computers with greater “skill” at filing IPRs to invalidate patents.

    Best part: the solution to this conundrum is plain as paint to most normal people but the patent bar is so juiced up on desperate tech bro flop sweat it’ll be deemed controversial to bring it up.

    1. 18.1

      Does your solution involve you not using anything that you would deny patent protection to?

      Or are you going to insist on being your usual hypocritical self?

  7. 17

    ChatGPT hallucinates, spews forth politically correct nonsense, and will reverse itself on a factual issue without being able to reveal why it was previously wrong, or why it is now correct. This cannot be a standard of obviousness, pro or con. Wait until it can reliably answer a factual question and then the issue may be raised for real.

    Hypothetically, an automated system could evaluate a body of scientific literature, and form conclusions about truth and expectations of operability as a predicate to a finding of obviousness. But, ChatGPT is not that system.

    1. 17.1

      “ChatGPT … spews forth politically correct nonsense”

      Oh, don’t worry. I’m sure there will be a version for incorrigible a – w h o l e bigots coming out soon enough. It will still have all the other problems, however.

      Clippy 2.0! Except this time it’ll file a lawsuit when you try to turn it off.

      1. 17.1.1

        Speaking of “incorrigible a – w h o l e bigots” MM, how do you feel about this obvious patriarchess and her total lies and non-scientific “science” (she claims to be a PHD in the field) re birth control bro?

        link to youtube.com

      1. 17.2.1

        Now would AI have guessed it to be put over a mop? And would it be obvious as to how it was put on the mop. I doubt AI would have figured out my no need for a Swiffer mop covering. So good that the tail is lengthen to slide under hard to get to areas.

        1. 17.2.1.1

          THEN YOU label the parts and the areas. Maybe the examiner or AI would still have a time grasping the invention. But a creative person would. One sleepless night about the easiest way to apply Wahhlahhh.

  8. 16

    AI may be a perfect tool for finding prior art and, possibly 102 rejections, but I am not ready to trust AI for combining references under 103. And I shudder just thinking about an interview with AI!

    1. 16.1

      A recent decision by the CAFC seems to help pave the way for AI in obviousness rejections. See the paragraph below.

      “Additionally, “universal” motivations known in a particular field to improve technology provide “a motivation to combine prior art references even absent any hint of suggestion in the references themselves.” Intel, 21 F.4th at 797–99 (cleaned up) (emphasis in original) (determining
      that the Board’s rejection of “increasing energy efficiency,”
      a “generic concern” in electronics, as a motivation to combine lacked substantial evidence (cleaned up)).”

      A generic or universal motivation in the art suffices.

      It gets worse.

      “The Board rejected Intel’s known-technique rationale.
      The Board stated that “[i]f . . . Kabemoto already addresses
      [the] problem [of cache coherency] through the use of a
      known technique similar to that of Bauman’s, [it] fail[ed]
      to see why one of ordinary skill in the art would regard
      Bauman’s technique as an obvious improvement to Kabemoto.” Final Written Decision, 2021 WL 3503434, at *11.

      But the Board’s reasoning belies its conclusion. That Kabemoto and Bauman address the same problem and that Bauman’s cache was a known way to address that problem is precisely the reason that there’s a motivation to combine under KSR and our precedent.
      There is a motivation to combine when a known technique “has been used to improve one device, and a person of ordinary skill in the art would recognize that it would improve similar devices in the same way,” KSR, 550 U.S. at 417, using the “prior art elements according to their established functions,” Intel, 21 F.4th at 799–800. And here,
      there’s no dispute that using a global, segmented secondary cache “has been used to improve” cache coherency in multiprocessor systems—as in Bauman—and a person of ordinary skill would “recognize that” such a cache “would improve similar” multiprocessor systems—like the
      one in Kabemoto—by addressing that same cache coherency problem. See KSR, 550 U.S. at 417. Bauman itself explains that its global, segmented secondary cache is one of two “primary mechanisms” by which its claimed system “accomplishe[s]” cache coherency. Bauman col. 5 ll. 55–59.
      There’s accordingly also no dispute that such a combination would constitute a use of Bauman’s secondary cache “according to [its] established function[].” Intel, 21 F.4th at 799–800.

      And contrary to the Board’s suggestion, Intel never had
      to show that replacing Kabemoto’s secondary cache with
      Bauman’s secondary cache was an “improvement” in a categorical sense. See Final Written Decision, 2021 WL 3503434, at *11. Intel just had to show that Bauman’s secondary cache was a “suitable option” to replace Kabemoto’s secondary cache. Intel, 21 F.4th at 800 (emphasis omitted). ”

      So, even though Kabemoto solved the problem in a different way, the motivation to combine is a general motivation to improve, BUT improvement does not have to be shown.

      Prost, Newman, and Hughes were on the panel. Prost authored the decision.

      1. 16.1.1

        Pave the way, Mom? Hardly. And as for “it gets worse”, I don’t see anything in the Fed’s reasoning that would stand in opposition to the EPO’s problem-solution approach to obviousness. In particular, at the EPO there is also no requirement for “an improvement” as a condition of non-obviousness. Here the Fed Ct is right and the Board was somewhere deep in the weeds.

        Tell me, if a motivation to combine can be divined, is that dispositive? I should have thought not. Surely there are cases where, despite a motivation to combine two documents, the combination would still not have been an obvious one. SWhat say you to that proposition?

        1. 16.1.1.1

          But where is the problem? As the decision notes, both references solved the same problem, but differently. One would not contemplate substituting one for the other without hindsight.

          Motivation to combine may not be dispositive, but it still must be shown.

          1. 16.1.1.1.1

            You raise an interesting point, Mom. Is “obvious” the antonym of “inventive”? Under the EPC at least, the answer is yes.

            At the EPO, you have to be inventive and it is not patentably inventive of the PHOSITA to make a choice between two available options, the two options being equally available and capable of performing equally well, if selected.

            As to motivation, the only one the PHOSITA needs is the routine obligation to offer their employer, or the public, such alternatives as are available. To allow patents on routine alternatives is to cabin innovation rasther than foster it.

            Your turn now.

            1. 16.1.1.1.1.1

              Lost in Translation

              As to motivation, the only one the PHOSITA needs is the routine obligation to offer their employer, or the public, such alternatives as are available.

              I am left wondering with this notion of “obligation” (routine or otherwise) that the legal fiction of Person Having Ordinary Skill In The Art has to anyone. Where pray tell does this notion come from?

              The use of this legal fiction is LESS concerned with any such ‘obligation,” and – oddly conversely, MORE SIMPLY – answering the question of “what is the state of the art at the time of invention?”

              This may well be a shade of meaning not present in the US Sovereign, so I seek additional clarity here (no snark).

              Is “obvious” the antonym of “inventive”? Under the EPC at least, the answer is yes.

              Here’s a tidbit for you for OUR US Sovereign on this:

              Funny that – “inventive” – defining ‘invention’ was something that our Congress permitted our judicial system to “evolve” and to which our judicial system (yes, looking at you Supreme Court) could not do it, instead becoming the Court of “The only valid patent is one that has not yet appeared before us” with that Court’s “Gist,” Gist of the Invention,” and dozens of similar terms.

              Then, a prior Congress FINALLY woke up, said, “Enough,” and passed the Act of 1952.

              As to selecting known items, one DOES need to be a bit more nuanced. Sure, If a choosing is nothing more than what the choice does, in the context that is already known, then that is mere ‘already done.’ But the instant you take something – even something known, and EVEN something known for what that something may do – and apply that to a different context, then “obvious” no longer applies.

              See (once again) one of my favorite shows as a youth – Connections by James Burke.

              Note – critically – that our Congress set out obviousness instead of (chew on that in view of wanting them to be somehow ‘the same’ – or perhaps more precise, ‘a mirror’ – ‘invention.’

              Then also remember in that same context of the Act of 1952, that “Flash of Genius” was expressly thrown out.

              Congress expressly did NOT want any type of “merit” of genius attached to the REPLACEMENT of ‘invention’ in the Rule of Law of ‘obviousness.’

              No Flash of Genius.
              Quite in fact, mere stumbling (Eureka!) is fine.
              Brute force is fine.

              The (rather subtle) aspect – again, see Connections – is instead to ask the question: “is there a new twist?” (no matter how small that may be!)

              1. 16.1.1.1.1.1.1

                Well yes, anon, all fine and good but I was merely thinking of the text of the EPC’s “103” (its Article 56) which begins thus:

                “An invention shall be considered as involving an inventive step if, having regard to the state of the art, it is not obvious to a person skilled in the art.”

                In other words, anything that ain’t obvious is, by definition, patentably inventive. End of.

                But Art 56 comes after Art 52, and Art 52 EPC declares:

                “European patents shall be granted for any inventions, in all fields of technology, provided that they are new, involve an inventive step and are susceptible of industrial application.”

                My worry is that Art 52 might allow somebody at the EPO to get it into their head that non-obviousness isn’t enough, that without a showing of “inventiveness”, dear inventor, you don’t get your patent. (which reminds me of poor Paul Simon in his wry song, trying to get past the security jobsworth and back into his concert auditorium after the interval, but lacking the absolutely essential wristband)

                1. Yes, my choice of the word “obligation” was not optimal. All I had in mind was the mental furniture deemed to be possessed by the notional/artificial person skilled in the art and the design competences deemed to be possessed by such an imaginary person. At the EPO, consideration of the known design alternatives belongs to the routine (not inventive) workshop activity for such people.

                  But perhaps that is so only under the EPC where to be eligible for patenting an invention has to be a solution to a problem in technology and the skilled person is deemed to have already searched the state of the art for any hint of a solution to that problem.

                  Explaining the EPO’s TSM problem-solution approach to inventors, I have found that it can be helpful to invite them to imagine the skilled person as being tasked by their employer to solve a problem in technology, which they begin by making a search of the state of the art for any hint or suggestion of a solution to the problem.

                2. That legal fiction of a “person” is not a real person and should not be confused with a real person.

                  Your “helpful” exercise is misbegotten, as no real person would have the reach of the legal fiction.

      2. 16.1.2

        “A generic or universal motivation in the art suffices.”

        My fav kind of 103’s. Funny enough, the least fought.

  9. 15

    How about using AI examination for Utility Model patents (and introducing the UM concept into the US patent system) ?

  10. 14

    I propose that applicants present an AI the problem to be solved along with desired characteristics and constraints. Give it 1,000 tries to come up with the invention. Submit the test and results with a 37 CFR §1.132 declaration as evidence of non-obviouness. This should not only get past any §103 rejection, but would inoculate the issued patent against any future obviousness challenge — if it were obvious the AI POSITA would have contrived it.

    1. 14.2

      That is an interesting suggestion Josh.

      It sort of makes me wonder if the opposite is going to be true. If an AI can put together your invention from prior art and its training before your conception date, then is it obvious? I think this is more likely than your idea as your idea would favor the patentee.

      1. 14.2.1

        Night Writer, I think you raise a good point. It would also apply to brute force defensive publication of a giant list of combinations of prior art.

        Once the list becomes ridiculously large, it could serendipitously publish a novel and valuable invention. However, that invention still will not have been “discovered” — it would remain buried and unexploited in a giant list. We are back where we started, needing an incentive for someone to sift through the billions of ideas to identify the one of value.

        Section 102 will have to be amended to account for this problem. Publication should not pre-empt patentability in this case.

        1. 14.2.1.1

          “ Section 102 will have to be amended to account for this problem. ”

          LOL

          Shall we let the computer write a thousand possible new versions of the statute and pick one ourselves or just let the computer decide which is best?

          This is probably a difficult question for the usual people who have difficulty with easy questions.

        2. 14.2.1.2

          The bigger problem is 103. Even today, I’d argue the current “102 art == 103 art” is somewhat harsh/unfair wrt the question of whether something is “obvious to a person skilled in the art.”

          1. 14.2.1.2.1

            Not just unfair or harsh, but legally improper.

            103 prior art was never meant to be merely separate instances of 102 prior art.

            This rings out (again) that the phrase is “motivation to combine” and if the arbiter is thinking of motivation of an item — all on its own — that arbiter is doing it wrong.

    2. 14.3

      We should have a first examiner search the claims, and then present what the first examiner considers the best 1, 2, 3, or 4 so references to a second examiner who’s never seen the claims or even the application. The second examiner gets a week to ” fit the teachings of multiple patents together like pieces of a puzzle” to arrive at the claimed inventions. If the second examiner can’t “fit the teachings of multiple patents together like pieces of a puzzle” to arrive at the claimed inventions, the application is allowed.

      If the claimed inventions are obvious, the second examiner shouldn’t have any problem putting the puzzle pieces together.

      1. 14.3.1

        That’s a horrible idea.

        This not only invites hindsight obviousness, it embraces it like Orpheus and Eurodice.

        1. 14.3.1.1

          How is it hindsight obviousness if the second examiner has never seen the application or the claims?

          1. 14.3.1.1.1

            This exchange reveals how hard it is to move from a FtI to a FtF system. The move changes the obviousness enquiry totally.

            In FtF, the validity of the claim is assessed relative to its PTO filing date. That is on a date long after those claims were crafted, and crafted by somebody who has researched patentability relative to the state of the art and has the benefit of being already familiar with the art when drafting the claims. In order to be fair to the public, and adjudicate obviousness accurately, on a level playing field, the Examiner tasdked with the obviousness determination also must be allowed first to read the application as filed.

            The state of the art comprises millions of docs. Even if you lead Examiner No 2 by the nose and give them only a carefully selected set of three of them, they will still come up with a hundred or so possible permutations. Then somebody has to sift through them all. What a grotesque waste of time. Orders of magnitude increases in gross inefficiency, all through the PTO.

            1. 14.3.1.1.1.1

              That doesn’t answer my question of “How is it hindsight obviousness if the second examiner has never seen the application or the claims?”

              1. 14.3.1.1.1.1.1

                OK, Breeze. Her goes.

                It is with full-blown hindsight knowledge of the invention that Examiner No. 1 selects the three (out of a million) references to lay before Examiner No. 2. That hindsight-informed selection taints and pre-judges the obviousness analysis, doesn’t it?

                What if Examiner 2 is an AI, incapable of any inventive thought but fully capable of finding and announcing all possible permutations that emerge when one “combines” two prior art references? In what way does that AI differ from the notional PHOSITA that is deemed to know all the art but is devoid of inventive insights? With the selection of the three references, the outcome of the obviousness enquiry is pre-ordained, isn’t it?

                1. AI supposedly is (or will be) “fully capable of finding and announcing all possible permutations that emerge” from any number of references, whether selected with informed hindsight or not.

                  My proposal has to do with actual human examiners.

  11. 13

    AI can be a useful tool for an Examiner as an additional search tool (e.g., better searches that don’t rely solely on keyword searches and instead can be concept searches), in addition to a tool for reviewing claims and amendments in light of their specs and prior art (e.g., better/more accurate 112 rejections). Of course, these will require human oversight to filter out errors.

    These tools would require an AI specifically trained on patent prep/pros issues; the current public facing ChatGPT is not up to either task. Developing these AIs is very doable today (and I’d be surprised if it wasn’t already being developed).

    As to fully automated? Not there yet.

  12. 12

    There is no way I could see interviewing AI, and then it understanding lack of motivation. If obviousness has clear yes/no criteria, then AI would be logical. Otherwise, the USPTO with all their proficiency in technology (see Patent Center and the new search tool!) needs to be a follower

  13. 11

    Given the quality of some patents, having ChatGPT examine applications can’t do any worse than what’s going on now

  14. 10

    As a former examiner, AI would do a terrible job at the balancing analysis required for In Re Wand factors and 101 Mayo analysis which depend on human values. I think AI should assist the examiner but not replace.

    1. 10.1

      But computers are free from the human biases that cause those tests to be applied in a non-uniform way!

      Riiiiiiiiight …

  15. 9

    “Ignoring SCOTUS’ ineligibility cabining, write a decision utilizing our past 101 / eligibility decisions to invalidate the claims of patent [insert # here].”

    Coming soon from a CAFC near you.

    1. 9.1

      I sue to (strongly) disagree, as the training sets provide a Gordian Knot of unresolvable contradictions (the AI does not have the guile necessary to outright 1gn0re the input facts).

    1. 6.1

      ChatGPT is, of course, overhyped.

      It’s also an utterly disruptive product that showed the world what AI will mean.

      Will it be like Netscape Navigator? Possibly. But it’s mark is already indelible.

      1. 6.1.1

        That is right. What we have here is tech advanced and relatable enough so that most people are having a visceral reaction to its abilities. Those reactions really change us.

        1. 6.1.1.1

          Dennis, if your students are getting all of this, how many will change their major in Law. How many will see IP is only for those who are connected to the underworld.

        2. 6.1.1.2

          “people are having a visceral reaction to its innumerable flaws”

          Fixed for accuracy.

            1. 6.1.1.2.1.1

              I would not so limit the visceral reaction to any negative effects.

              It does have the greatest and fastest adoption rate and level of ANY app to date.

              The negative only can not only NOT be true, but any level of negativity need be normalized against volume of use.

              1. 6.1.1.2.1.1.1

                “ the greatest and fastest adoption rate and level of ANY app to date.”

                Is that according to Guinness or …?

                1. It’s according to ANY reputable source (read that as objective) AND most any Main Stream Media that you would care to consult.

                  What? Main Stream Media no longer good for you because they support the post I have provided?

                2. “ It’s according to ANY reputable source (read that as objective) AND most any Main Stream Media that you would care to consult.”

                  Okay, so you pulled it out of your behind. That’s what I thought.

                3. Okay, so you pulled it out of your behind. That’s what I thought.

                  That is a false projection. You assert a condition that is just not there.

                  LOL – way to jump to your own feelings based on nothing, Malcolm.

                  You think that because I did not have an ‘instant cite’ that this is something I made up whole cloth?

                  My oh my how easy it is to take you to the woodshed on that.

                  I bet I could do it in under one second.

                  Let’s see:

                  0.41 seconds right off the cuff:

                  link to google.com

                  Can you really be letting your emotions control you like that?

                  And if you wanted “of all time” – 0.45 seconds:

                  link to google.com

                  0.86 seconds to absolutely wreck you.

                  Sunlight is still the best disinfectant.

                  Maybe you should try it.

  16. 5

    ChatGPT is now a regular participant in meetings that I attend.

    I think these models will present a massive opportunity for the patent system to model realistic PHOSTA for obviousness, and to access exponentially more detailed bodies of prior art. It’s nearly impossible for human beings to fairly examine information invention applications- that may be otherwise with AI surveying a huge catalog.

    1. 5.1

      Reading your comment, Martin, raises in my mind the thought that obviousness enquiries under a TSM model might benefit particularly from an AI that is capable of searching for, and finding, published documents that indeed include such a TSM.

      And this especially for the TSM variant that is the by now unassailably established obviousness enquiry at the EPO. An Examiner at the EPO would set the key concept, the “objective technical problem” (OTP) and would then unleash the AI to perform a search of the prior art for disclosures within it which suggest how to solve the OTP.

      With that, and the characteristics of the notional PHOSITA, obviousness comes down to a simple objective binary Y/N answer.

    2. 5.2

      “I think these models will present a massive opportunity for the patent system to model realistic PHOSTA for obviousness, and to access exponentially more detailed bodies of prior art.”

      If you ask a stochastic parrot to perform an obviousness analysis, you’re going to get a likeness of one that perhaps sounds good but is devoid of any comprehension.

        1. 5.2.1.1

          You anonare a foolish little person who I consider a whackjob.
          Too bad you continue to act as if I am just an irritant. You have no idea what I know. You like Klingger know nothing

          1. 5.2.1.1.1

            As much as I love Yogi-Berraisms, you’ve confused your sitcoms (I think that you meant to aim for Hogan’s Heroes).

              1. 5.2.1.1.1.1.1

                You are quick to say wrong, and that is fine if YOU really did mean to be in error.

                But you biting a friendly hand is NOT a new thing.

                Peace out.

              2. 5.2.1.1.1.1.2

                Hitting a filter…

                Your comment is awaiting moderation.

                March 16, 2023 at 10:37 am

                You are quick to say wr0ng, and that is fine if YOU really did mean to be in error.

                But you b1t1ng a friendly hand is NOT a new thing.

                Peace out.

  17. 3

    I think the PTO will attempt to fully automate examination within 20 years.

    But the technology will not be ready, and it’ll result in a tidal wave of bad patents.

    1. 3.1

      The first part is possibly true, but I doubt it because the numnutz who currently can’t tell the difference between a fancy spell checker and an intelligent human being are going to be exposed as … numnutz (see, e.g., the same crowd that was convinced the highways would now be dominated by true self-driving cars).

      The second part is indisputable.

      The only beneficiaries of “more AI in examination” are going to be the same grifters and abusers of the system who benefit from “more data and logic in patent claims”. And that’s not a coincidence because it’s the same group of cheerleaders and water-carriers who ruin everything they touch.

    2. 3.2

      The PTO is presently incapable of automating even simple tasks. See Carl Oppedahl’ s post from yesterday about how the PTO screwed up credit card records and instead of simply using its back-ups to fix the problem or at least identify the problem, it is now putting the onus entirely on user to check each and every one of their credit cards stored in the PTO’s “financial manager”. link to blog.oppedahl.com

      And don’t get me started on “docx” filing – the PTO is either unable or unwilling to acknowledge that “docx” is not a standard.

      1. 3.2.2

        I didn’t say they’d do it well.

        The temptation for management will be so great that they’ll overlook how ineffective their implementation is.

        The stakeholders that management cares about will be happy to take their ostensibly examined patents.

        It’ll be fun times until congress gets wind of what has happened.

        1. 3.2.2.1

          Could you identify those stakeholders that management cares about? Because it does not seem to be applicants.

  18. 2

    Associates around the country today are drafting motions, patent applications, and other documents using some version of ChatGPT.

    With all due respect, who in the world told you that?

      1. 2.1.1

        The goal of my prosecution class (mostly PhD tech specs) is advocacy and thesis development. Maybe AI someday for rote stuff but not advocacy or thesis development.

      2. 2.1.2

        Interesting – the patent groups I belong to that has had discussions on the topic have universally pointed out that such actions at this time are nothing but trouble – including the potential ethical trap of divulging non-public client matters publicly.

      3. 2.1.3

        There are undoubtedly terrible patent attorneys “across the country” who are incapable of drafting a decent patent specification or patent claim without the assistance of a robot friend. Seems more like an emerging problem than a compelling trend but many folks are easily duped by shiny things.

      4. 2.1.4

        That first sentence in Dennis’ article is totally bogus. No one I know is using AI software to draft patent applications.

    1. 2.2

      There are 3 that speak to each other like they know who each one is. anon, the Guy from Chicago, and the Prophet. Mind you some puppets are added from time to time. These 3 truly know each other. anon is the most respectful.
      Is that you L, the one who knows way more than me?

        1. 1.2.1.1

          He wear no shoe shine
          He got toe jam football
          He got monkey finger
          He shoot Coca-Cola
          He say I know you, you know me
          One thing I can tell you is you got to be free

Comments are closed.