Commentary

Can an artificial intelligence model be the inventor of a molecule designed by the model and how can patentability be assessed?

Michael Huhn

 

1 Introduction

 

    It appears not  appropriate to refuse  patentability of an inventionon a newmoleculede-signed  by  AI  because  the  respective  patent  application  does  not  have  significant  examples which were carried out in reality,  but only generated  by  AI  (usually  a  trained  model/machine trained algorithm). However, to achieve patentability,  certain  requirements  must  be  fulfilled, in particular relative to the estimation accuracy of  the  trained  model  and  to  successful  repetition of the examples  in view of known state  of the art at the filing date.

    It  is  more  than  questionable  if  an  AI  model can  be  the  inventor  of  a  molecule  designed  by the  model,  the  first  patent  applications  in  this respect  having  been  filed.  Assessment  of  inventive step for new molecules generated by AI should  remain  subject  of  discussion.  There  are no  clear  positions  by  the  patent  offices  for  the time being.

    It  is  a  well-known  fact  that  Artificial  Intelligence  (AI)  has  a  vastly  growing  impact  on  our everyday  life,  for  carrying  out  innovative  and creative   acts   resulting   in   inventions,   which could previously only be made by humans. A rapid development in the use of computers in chemistry  could  be  observed  in  the  last  50 years. A further immense development, now in the  design  of  new  molecules,  occurred  when computers became powerful enough to process machine learning algorithms, discover patterns in  data,  and  construct  mathematical  models using these discoveries. Algorithms can be provided  with  data  to  learn  from  (trained  model). This  is  the  principle  of  Artificial  Intelligence. Patents can be issued on trained models themselves, or the trained model can be applied, e.g. in  pattern  or  language  recognition  or  the  subject-matter  of  the  present  article –designing molecules  (Chen  et  al.,  2018;  Engkvist  et  al., 2018; Sellwood et al, 2018).

    Many questions relating to the protection of these  inventions  yet  remain  to  be  answered. The  readers’ attention  is  hereby  drawn  to  the fact  that  the  field “AI  and  patents” is  still  in  a very   early   stage   of   development.   Little   is known.  Only  a  limited  number  of  publications related  to  this  subject-matter  exist,  and  it  cannot be said that it is easy to get an overview on the actual state of art in the field. Some articles are  very  general,  some  treat  the  use  of  AI  in drug  discovery, some  give general overviews of the various fields where AI can be applied, some disclose in which technical fields AI is used, and so forth.

 

    The following publications are of considerable interest in the present field:

“Artificial     Intelligence     and     Drug     Discovery” (Leanse, T., 2019)

 

“Artificial  intelligence:  the  implications  for  patents” (Kuhnen, R. K., 2019)

 

“Artificial  creativity—is  the  IP  system  ready  for robot inventors?” (Inchley, T., 2019)

 

„Machine  yearning:  AI  and  patents” (various authors, 2019)

 

“Patenting Artificial Intelligence: Issues of Obviousness,    Inventorship    and    Patent    Eligibil-ity“ (Tull, S. Y. and Miller, P. E., 2018),

 

“WIPO Technology Trends 2019 Artificial Intelligence” (WIPO, 2019).

 

    A frequently encountered question concerns not  the  patentability  of  an  AI  method  as  such, but  of  molecules,  materials,  compositions  and the  like,  designed  (conceived)  thereby.  In  these cases,  the  human  (i.e.  the „classical“ inventor) plays a lesser and lesser role. It is expected that this  will  have  an  impact  on  the  assessment  if results  (examples)  conceived  by  AI  meet  the requirements for sufficiency of disclosure.

    It  is  assumed  that  AI  is  used  frequently  in chemical  and  pharmaceutical  industry  to  design  new  molecules  or  related  compositions  of matter. However, it is not clear to which extent AI  is  used  since  industry  is  rather  silent  in  this respect.  Furthermore,  the  number  of  filed  patent applications cannot be taken  as an indicator.  Due  to  uncertainty  if  protection  of  a  molecule designed by AI (and not in the lab) is avail-ableatall,industryhasnotfiledpatentapplications in this field. In surplus, the questions who is  the  inventor  of  the  molecule  and  how  the inventive  step  (i.e.  if  the  new  molecule  is  sufficiently  distinct  from  the  prior  art)  is  assessed are not clear. As long as this is the case, patent applications will not be filed.

    Hereinafter, it will be tried to give an answer to the question if it  is or will  be  possible to  patent   molecules,   materials,   compositions   and the    like    showing    advantageous    properties which  are  designed  by  AI  and,  in  the  affirma-tive,  if  the  AI  model  is  the  inventor  of  the  new compound.  The  author  will  furthermore  address  some  crucial  questions  relating  to  the assessment of the inventive step.


2 Patentability of Molecules designed by Artificial Intelligence


    In the present context, molecules having pharmacological activity (interaction with tar-gets, e.g. antigens, antibodies, enzymes) play a paramount role. However, the results provided below also apply for materials, compositions and the like not having a physiological, but other activity.

    In the context of the present article, the term “molecule” refers not only to molecules, but also to materials, compositions and the like including DNA, enzymes, antibodies, (liquid) crystals, just to name a few.

    The interaction of molecules with certain targets can be calculated very accurately today using AI. Even though this is nothing else, in principle, than well-known „in silico chemistry“, calculations supported by AI (“trained model”) now have a more accurate scientific basis, generating in many cases precise results in shorter time.

    In consequence, an actual question in this respect is whether „AI-generated“ (“trained model generated”) molecules having certain (alleged) properties can be patented as such, even though they were not synthesized and tested in vitro at the priority date. 

 

    To answer this question, the two decisive questions criteria should be:

 

1) Does a patent application on a molecule generated by AI provide ample disclosure in the description and the (not real) examples for the person skilled in the art to enable synthesis of the respective molecule in vitro?

 

2) Does the skilled person, at the priority date, assess the examples (and the respective parts of the description) as credible, because they do not contradict common teachings and/or the estimation accuracy of the trained AI model is sufficiently high?

 

    It is held that application of the above two criteria could serve to avoid that the examples in the respective patent application are just an (uneducated) guess not having a sound scien-tific basis (meaning that even if the examples of the application could be successfully repro-duced, this was purely accidental).

    The above approach is supported, on the one hand side, by the Japanese Patent Office JPO, in "Examination Guidelines for Patent and Utility Model" (JPO, 2019a), „Case examples pertinent to AI-related technology” (JPO, 2019b) and „Newly Added Case Examples for AI-Related Technologies“ (JPO, 2019c) (Presentation Material).

    Example 51 in “Case examples pertinent to AI-related technology” and „Newly added case examples for AI-related technologies” is a fictive example for a patent application not providing enabling disclosure. The application is on a curable adhesive invented by a trained AI model. The adhesive has a certain composition to cure faster than state of the art adhesives. No real examples are found in the de-scription, only an example created by the trained model. The estimation accuracy of the trained model has not been verified.

    The facts that a) it was common technical knowledge at the priority date that it is difficult to control the curing reaction the way de-scribed in the patent application; and b) the example is a “trained model example” created without a verified estimation accuracy, are rea-sons that the application is assessed as not providing enabling disclosure (written support) in the description. This cannot be remedied by later filing data showing that the trained model result was correct, as the skilled person would not have believed that the claimed invention can be carried out at the priority date, for being a) contrary to common knowledge and b) based on speculation. However, this conversely should mean that the invention would have been patentable if the two above criteria had been met.

    The actual „Guidelines for Examination“ of the European Patent Office EPO answer almost exclusively questions related to assessment of inventive step and technicity of AI methods (Guidelines for Examination in the European Patent Office, November 2019, Section G-II, 3.3.1, Section G-II, 3.6, G-VII, 5.4) (EPO, 2019). Unfortu-nately, support for the correctness of the above assessment is not found there.

    Such support, however, appears to exist in case law of the Boards of Appeal of the EPO. It is pointed to case law on so-called „prophetic ex-amples“, which are established as proof to show that an invention can be carried out at the priority date. Definite proof can then be filed at later points in time by “real” examples. However, in general such proof is only accepted if the teachings of the claims and the descrip-tion is not contrary to the general teachings in the particular field at the priority date. Deci-sions have to be taken on a case-by-case basis.

    In the present context, the decision T2220/14 (EPO Boards of Appeal, 2015) backed up by T1496/08 (EPO Boards of Appeal, 2012), is worthwhile mentioning.

    T1496/08 states the following (p. 20, 1st par-agraph): “Post-published evidence may be taken into account, but only to back-up the findings in the patent application in relation to the use of the ingredient as a pharmaceutical, and not to establish sufficiency of disclosure on its own.” T2220/14 states the following in Point 63.of “The Reasons for the Decision”: The respond-ents have not presented convincing evidence that this would be the case, their main argu-ment being that Example 3 is a "prophetic“ ex-ample. However, there is no requirement in the EPC that, either at the priority or filing date, the applicant must have carried out the claimed invention. The requirement of Article 83 EPC is that a person skilled in the art, following the teachings in the application as filed supple-mented with his/her common general knowledge and with a reasonable amount of experimentation, including some trial and error,would be able to carry out the invention as claimed at the relevant date. (emphasis added).

 

3 Inventor questions


    Another crucial question is: who is the in-ventor of molecules designed by AI? The person who has created the trained model and/or who has applied the trained model to find the new compounds? According to generally applied principles, an inventor must be a natural person (it should be noted, however, that this is not explicitly required by the European Patent Con-vention). However -what to do when an inven-tion has been clearly made by a machine trained algorithm? Until now, for “serious” in-ventions having a potential commercial value, no one will name the trained model as an in-ventor, because it seems clear that the applica-tion will be rejected for not complying with inventor requirements.
    However, recently two patent applications were filed in various countries by the same applicant (Dr. Stefan Thaler) which have in the meantime been published under the numbers EP 3 564 144 und EP 3 563 896 by the EPO . A machine trained algorithm was named as the inventor. The algorithm as such appears to be protected by a patent application (US 2015/0379394), naming Dr. Stephan Thaler as an inventor. The subject-matter of the patent applications are a food container and an elec-tronic device.
    More information is available on the web-site of the EPO (EPO, 2019), the magazines “The IPKat” (Hughes, 2019a, 2019b; Papadopoulou, 2019) and “iam” (Wild, 2019). This case should be a “trial balloon” challenging the Patent Offices to give an answer to the crucial ques-tion if a trained model can qualify as an inven-tor.

 

4 Inventive Step

 

    A further, important topic, frequently also encountered when molecules are designed by AI is the inventive step. “Inventive Step” or also “Obviousness” refers to the patentability crite-rion if the new invention is sufficiently remote and different from what is known in the art (the pool of publications in the same field) is not considered “trivial”. 
    Let’s take the case that an individual helps to create a trained AI model/machine trained algorithm. The model reveals to give excellent results in designing molecules having certain desired properties, e.g. binding to certain tar-gets (e.g. enzymes, receptors in the medical field) or lending themselves as perfuming in-gredients, colorants or sweeteners, just to name a few. The person having conceived the trained model is the inventor, in the classical sense, of the model; but also (very probably, see below) of the new molecule. Until here, the sto-ry is still easy. But how about the assessment of the inventive step if the same model is used again to design further molecules? It appears that the threshold for patentability relative to inventive step becomes higher, or that the inventive step will be even denied. The design of a new molecule using the same model which has been already successfully applied in the design of the first molecule could be regarded as a simple routine act, even though the spe-cific molecule provides advantageous proper-ties and would be regarded as inventive under “classical” criteria.
    It is not clear if one day the respective Pa-tent Offices will take the above approach. In any case, applicants wishing to patent new molecules designed by AI may prefer not to dis-close that the “method behind” the creation of the new compounds is a machine trained algo-rithm, in order not to “raise the bar” for the in-ventive step or, rather, to have the examiner apply the “classical” criteria. Applicants may even think of not disclosing the model in the first application (in which the trained model was used for the first time), i.e. not to mention the model. In this respect, however, the ques-tion arises if it does not become evident that the new molecules were the result of a trained model and the application gets rejected for lack of disclosure. The AI used to design new mole-cules should in principle be open to protection by patents. Such exclusivity for the best AI would clearly provide the company, often a drug company, with a competitive advantage. However, the protection for molecules, and in particular the important “crown jewels”, might get lost, in a worst-case scenario.

 

5 Conclusion

 

    As more and more new molecules are de-signed by AI, without any examples having been carried out in vitro, the question arises if the design of the particular molecule results in a patentable invention. To answer this ques-tion, it appears appropriate to use argumenta-tion based on patents having prophetical ex-amples (the other type of examples which have not been carried out when the patent applica-tion was filed). Since patents having prophetical examples can be granted under certain conditions, this should also apply for patents having only examples for AI-designed mole-cules. A careful analysis of the Guidelines for Examination of the JPO and the above-cited EPO case law (EPO, Guidelines for Examiniation) reveals which criteria for patentability should be checked if the examples can be successfully carried out. It should be verified if the examples are in line with common knowledge at the pri-ority date and if the estimation accuracy of the trained model is sufficiently high. Then the success was not accidental.
    It seems that the bar for patentability of compounds designed by AI will inevitably be raised. Many questions cannot be answered for the time being, one of them being if AI pro-grams can be inventors. Two cases are known to date in which patent applications naming an AI inventor have been rejected by the European Patent Office (decision can be appealed). As the reasoning for the decisions is not available yet, it is not clear what is behind the decision, but it is assumed that the EPO will base it on the rea-son that the inventor is not a human being.
    It is not clear how the inventive step will be assessed in case an AI-designed compound was found patentable and the same trained model shall be used again to design a (further) com-pound. In such a case, the examiner may argue it was known that the trained model is capable of successfully designing new molecules with some desired properties. The design of another molecule will then just be the result of a rou-tine act, namely providing the relevant data to the model. At present it is not clear how such an objection can be avoided or overcome. One solution might be not to disclose that the mole-cule was designed by a machine trained algo-rithm. This should avoid the objection that the new molecule was created in a routine act. However, shouldn’t it immediately become ob-vious that the examples are only based on AI? Maybe this does not even trigger negative con-sequences as after all the situation appears very similar to a “classical” pharmaceutical pa-tent application with prophetic examples (which would correspond to the AI-generated examples). Yet this would mean that the sub-ject-matter of the application is patentable!

 

 



____________________________________________________________________________________________

 

References

 

Chen, H., Engkvist, O., Wang, Y.,  Olivecrona, M.  and  Blaschke,  T.,  (2018):  The  rise  of  deep learning  in  drug  discovery, Drug  Discovery  To-day, 23(6), p. 1241-1250.

 

Engkvist, O., Norrby, PO., Selmi, N., Lam, YH., Peng,  Z.,  Sherer,  E.  C.,  Amberg,  W., 

 

Erhard,  T.  and  Smyth,  L.  A.,  (2018):  Computational  predic-tion  of  chemical  reactions:  current  status  and outlook., Drug  Discovery  Today, 23(6),  p.  1203-1218.

 

EPO, Guidelines  for  Examination,  Section  G-II,  3.3.1,  Section  G-II,  3.6,  G-VII,  5.4,  available  at  https://www.epo.org/law-practice/legal-texts/html/guidelines/e/index.htm, accessed 17 Janu-ary 2020.

 

EPO   Boards   of   Appeal,   (2015): T   2220/14 (VelocImmune  mouse/REGENERON), available at https://www.epo.org/law-practice/case-law-appeals/pdf/t142220eu1.pdf accessed 17 January 2020.

 

EPO   Boards   of   Appeal   (2012): T   1496/08 (Therapeutic  vaccination/BN  IMMUNOTHERA-PEUTICS, INC.), available at https://www.epo.org/law-practice/case-law-appeals/pdf/t081496eu1.pdf, accessed 17 January 2020.

 

EPO,   (2019):   EP3563896 -DEVICES   AND METHODS   FOR   ATTRACTING   ENHANCED   AT-TENTION,available  at  https://register.epo.org/app-lication?number=EP18275174&lng=en&tab =main, accessed 17 January 2020.

 

EPO,  (2019): EP3564144 -FOOD  CONTAINER, available at https://register.epo.org/application?number=EP18275163&lng=en&tab =main, accessed 17 January 2020.

 

Hughes,  R.,  (2019a): The  first  AI  inventor -IPKat  searches  for  the  facts  behind  the  hype, available   at   http://ipkitten.blogspot.com/   20 19/08/the-first-ai-inventor-ipkat-searches.html, accessed 17 January 2020.

 

Hughes, R. (2019b): EPO refuses "AI inventor" applications in short order -AI Inventor team intend to appeal, available at http://ipkitten.blogspot.com/2019/12/epo-refuses-ai-inventor-applications-in.html, accessed 17 Janu-ary 2020.

 

JPO (2019a): Examination Guidelines, availa-ble at https://www.jpo.go.jp/e/system/laws/rule/guideline/patent/ai_jirei_e.html, accessed 17 January 2020.

 

JPO (2019b): Case examples pertinent to AI-related technology,available at: https://www. jpo.go.jp/e/system/laws/rule/guideline/patent/document/ai_jirei_e/jirei_e.pdf, ac-cessed 17 January 2020.

 

JPO (2019c): Newly added case examples for AI-related technologies, available at https://www.jpo.go.jp/e/system/laws/rule/guideline/patent/document/ai_jirei_e/jirei_tsuika_e.pdf, accessed 17 January 2020.

 

Inchley, T., (2019): Artificial creativity—is the IP system ready for robot inventors?, available at https://www.twobirds.com/~/media/pdfs/artificial-creativity.pdf?la=en&hash=580937B3 6F70219EB20CE4D728008D76B622A918, ac-cessed 17 January 2020.

 

Kuhnen, R. K., (2019): Artificial intelligence: the implications for patents, available at https://www.iam-media.com/artificial-intelligence-implications-patents, accessed 17 January 2020.

 

Leanse, T., (2019): Artificial Intelligence and Drug Discovery, available at http://www. mon-daq.com/uk/x/805036/Patent/Artificial+ Intel-ligence+and+Drug+Discovery, accessed 17 Janu-ary 2020.

 

Papadopoulou, F., (2019): Inventorship under the light of AI?, available at http://ipkitten. blogspot.com/2019/11/inventorship-under-light-of-ai.html, accessed 17 January 2020.

 

Sellwood, M. A., Ahmed, M., Segler, M. HS., & Brown, N., (2018): Artificial intelligence in drug discovery, Future Medicinal Chemistry, 10(17), p. 2025-2028.

 

Tull, S. Y., Miller and P. E., (2018): Patenting Artificial Intelligence: Issues of Obviousness, Inventorship, and Patent Eligibility, RAIL The Journal of Robotics, Artificial Intelligence and Law,1(5), p 313-325

 

Various authors (2019): Machine yearning: AI and patents, available at https://www. man-agingip.com/pdfs/Machine-yearning-AI-and-patents.pdf, accessed 17 January 2020.

 

Wild, J., (2019): EPO says no to patent appli-cations for inventions made by machines, avail-able at https://www.iam-media.com/law-policy/epo-turns-down-ai-application, accessed 17 January 2020.

 

WIPO, (2019): WIPO Technology Trends 2019 Artificial Intelligence, https://www.wipo.int/edocs/pubdocs/en/wipo_pub_105


schatten_unten