[[These transcripts are generated automatically by WIPO Speech-to-Text© from the audiovisual recording. All non-English language versions of the transcripts are generated automatically from the English text by WIPO Translate©. The accuracy of the transcripts and their translations cannot therefore be guaranteed. Only the original speech of the audiovisual recording constitutes the official record of the meeting proceedings. ]]
Good morning, everyone.
The next item is the vote.
Good morning, everyone, and welcome back to our session this morning, JIPA had a lovely evening.
Thank you,
So we will be having some further very interesting discussions today and this morning we will continue our discussions on quality of patents, including opposition systems.
So to begin with what we will do is move to the sharing session on the use of artificial intelligence for the examination of patent applications.
So I will open the floor and invite delegations to share information on their use of artificial intelligence for that purpose, will begin with presentations.
We are using using the screens so we have Japan the United Kingdom France.
Spain and the presentation by WIPO and then I will open the floor for further interventions so if to start with, I can invite the Japanese Delegation.
Thank you,
Thank you.
Thank you,
Thank you,
Thank you.
Good morning, everybody, my name is Igataka Hiro, Assistant Director of the International Policy Division, the Japan Patent Office.
I would like to thank you for giving me the opportunity to show you the JPO's initiative for using artificial intelligence in our business operations.
In this right, I will talk about the purpose of using AIEA's JPO's business operations.
In recent years, the cutting age technologies have been advanced and complex, examiners need to deal with such joint.
And also due to the growing numbers of applications field worldwide, the scope of prior art searches has been expanding.
In addition, the number of trademark applications has been underlies in recent years.
The examiners need to conduct trademark examinations without delay.
It is also an essential issue for the JPO to find ways to pass on the expertise on complicated operations from experienced ourselves to new officers.
Under these circumstances, we find lapping advances in technologies such as machine running, Newland, Dongage, processing, and deep learning,
Some of these technologies are expected to be used in our business operations.
Based on these, the JPO that started to consider the best possible use of AI technologies for its business operations.
Our ultimate goals of using AItechnology eats.
It is to solve various issues that JPO should deal with so as to conduct higher quality and more efficient business operations.
As a result of this, we hope to improve our services to users and reform works time conducted by other colleagues.
This right shows the progress of the JPO's initiative to make use of AIC.
In 5, 2017 we studied possibilities of AItechnology Africa soon to our business operations.
After reviewing our entire business operations we have selected the operations where air technology can be possibly applied.
In FI 2018 we have conducted proof of concert POC and also done the clarification of presideous risks.
5, the selected operations.
For FI 2019 we have started preparing the development of the support tools for JPO's internal operations along with the communication of
The continuation of conducting PLC for some of the certain operations.
JPO has considered the application of AIE in accordance with action plan since 2005.
The action plan was revised in 2018 based on the result of the consideration in FI2017.
In July 2018, we conducted three projects to prepare for the introduction.
Three projects are sounded by green line in discharge.
These are response to inquiries, trademark image, such and classifications of goods or services for trademarks.
Especially for the PLC project on response to inquiries.
We have created learning data to build assistance, that support speaking word to text transcription, using post-recognition.
Further, since by 2018, we have started PLT project, not only on patent e-as such, but also on
There is an e-mausage with a focus on preinal designs.
Also not rested in this table for FI 2019 with new re-study two PLT projects which are on.
Such lasart, linking and summarization.
While POC project on such result ranking is aimed to sort much through a patent document according to the similarity to the claimed invention.
The one on the summarization is aimed to extract broad extract programs, purposes and effects from the description of patent documents.
Though we do not really have enough data or PLT without yet, we are moving forward to make
In preparation to develop support tools for JPO operations to accelerate the implementation of the AI technology.
AIBS too had been developed by e-hasty using an agile software development approach.
In this way, we can catch up with lapidary developing AI technology and incorporate the latest technology into the two.
By shortening the user, PDCA, BRAM do check act cycle with other software development approach, we can let some use.
Users use the two more quickly and therefore we can expect to improve the two functions further based on the user fee back.
The reason why we made enhanced development team to create an eye-based to avoid Aids Broadbox program.
We believe that the SCP is suitable for them to share our views and experiences.
We look forward to further discussion at this Forum, thank you very much for your kind attention.
Thank you very much for that presentation, next I have the UK.
Thank you,
Thank you,
Thank you,
Good morning, everyone, from the Intellectual Property Office of the UK, so just before starting, I would like to thank the Delegation of Japan for
The very informative presentation on the use of AI within the patent examination process and other initiatives that they have going on there.
We are also looking forward to the presentations or statements that is still to be made by other colleagues later this morning.
It is very likely that such tools if designed and implemented correctly have the potential to boost examiner efficiency as well as patent quality which are laudable benefits.
The UK hopes this opportunity to share experiences will not only increase awareness of ongoing projects will avoid but potentially enable better development of such tools and create opportunities for collaboration on this work.
The UK itself is still in the early stages of exploring and developing AItools for you since patent examination but I will share with you today an overview of our progress so far.
So the UK Government provides a what is called regulator's pioneer fund to help address barriers to innovation in the UK economy.
Under this the IPO was one of thirteen UK regulators to secure funding and using it to explore A solutions that could enhance and modernize the online filing process for IP rights.
Thank you,
Within this scheme, the IPO are currently running two projects with this funding, so the first is a research study to understand the feasibility, technical complexities, and effectiveness of AI assisted
Patent prior art searching, the second is the creation of a production system for utilizing AIfor recognition of earlier trademarks and design rights.
So obviously given the mandate of this committee and of this sharing session, I will just go on to detail the project on patent prior art searching here, so in January 2019,
A local university card of university one the contract out of the funding to undertake a feasibility research study on how AI could assist patent examiners as part of the prior art search process.
The aim is to run an A Iowagratham on a patent application for the computer to read the text scan the worldwide patent databases and determine the top fifty or so documents that it considers to be the most relevant to their invention.
These documents would then act as a strong starting point for an examiner search identifying potentially relevant citations and making the prior art process more time efficient.
The keeper of the hour at them developed to date involves keeping the human in the Luke as part of a human machine system this means using the expertise and experience of patent examiners to bolster any automated search query to.
Result in the benefit of a kind of enhanced brain to achieve better results so three technology sectors are being focused on for the purposes of this feasibility study.
And being computing, civil engineering and transport, these were chosen because they are the IPO three largest sectors in terms of volume of searches over the past decade, so it was a sensible place to start.
I will go with them testing.
With our patent examiners commenced in October 2019 so very recently and this involves specific tests to evaluate the quality of the topic modelling alrhythms as well as information retrieval testing.
To assess the relevance of the AIsearch results. The results of this testing are now being written up and a research paper will be presented to the Intellectual Property office later this month highlighting outcomes of this feasibility study.
We will then consider the next steps to this work.
So if any of my colleagues here today would like any further information on our work in this area, I would be happy to take away any queries you may have and to go forward and share any future experiences on the work you are doing and we are doing
I would like to invite the Delegation of France.
Thank you,
Thank you,
Thank you.
Good morning, everyone, I am Jonathan Witz.
Here at the SCP for a few years now you should know my face I would like to present the first experience of the MPA in the area of AI this is a project which aims to improve her.
The processing of applications, so why are we doing this, I want to put your efficiency
How to do this while we are going to build a proof of concept to assess the feasibility of our ideas.
And if the results are OK we will move on to the industrialization phase and roll out in various services otherwise we will use this experience.
To capitalize on our knowledge and see what we can do when we build new tools next time.
What is actually going on at the INP, we have four and 15 new applications per week and there dispatch to three services, there is a service,
Which deals with the applications that deal with mechanical issues, other chemical issues and third a third service that has to do with the new technologies.
Moving on the services dispatched their results to sub services for the mechanical service, I would have a service that would deal with transport issues.
Ah perhaps another would be buildings and so on and so forth the same thing goes on for chemical the chemical despattress is organical chemistry and so forth.
So we call them P eleven P twelve be thirteen and so on and so forth them now this dispatching of the request by managers that is the people who are responsible.
Four services is one as to as three as is a quite a long it is quite a long time takes the nine hours per week and then the sub service managers.
I use up to two hours pertain per week for a total of twenty hours to dispatch the requests applications rather spatching.
Was based on the texts of the applications and their claims.
So the idea was to creator a key to despatch things to save these nine hours per week for the service managers and the two hours for team.
So that is I came to a total of about twenty hours or week.
Here are the key exhortations as I have said we have an operational name as to have a tool.
That would propose classification and would automatically dispatch the applications to the appropriate teams.
Here is a slide on organization, we have the Department of Patent Department.
Then we have a
On research laboratory that is part of the NP and green you have the IT department which is the support organization for the practical implementation of tools as created by the laboratory.
And obviously the department of patents validates the use of tools make suggestions to improve these tools.
Now we can talk about insertion of the AI model in the patent process.
The lab's proposal was to come up with the classification model which was based on machine learning and which.
Would allow us to have files that would be part or integrated in the patent back office processing.
This man that would we would have fee dispatching mechanisms that.
The various subservices would see the documents automatically and can see that the.
Patents are put in a specific repose story go through the black box the prediction model and these lists are reinjected in the IT systems towards the department.
The patent apartment and towards the patent examiners here's a slide on building the machine learning model.
Oh I am not an IT expert so I can not explain this in any detail having said that the principle is to have this we collect.
Data, this is open access data on patent applications, we collect the
Publication numbers and there I am P classification and we asked the machine to make a link between the text and the classification that it made and so.
And this a louses for the IE to be come involved.
Then we link this knowledge with our and dispatching by area of expertise and depending on the text of the request that allows us to analyze the text.
To have a proposal for classification and then to send on the information to the appropriate examination area using artificial intelligence.
Thank you,
Now looking at the performance of the model currently we have dried it eighty five per cent accuracy eighty five per cent of the time I.
The time at Sir the eye made the right classification eight per cent was ambiguous and other areas were clear error.
If you look at your performance of around 90% and you, you can see that the A, accuracy is fairly good already.
This is more a technical slide so let me go over rather quickly we have our shaves that are unsipped and then.
They are prepared so that the model can read them easily.
I am sorry, these are texts.
Files and the application reads them and makes protection on the basis of its analyst analysis of the information.
Based on existing knowledge they are able to make an approximation approximate analysis and recommendation based on the information in the textiles.
His line that covers the results we saved five or six.
I was per week for the service managers or yes one as to us three level managers and we saved one or two hours per week for each.
Each a sub manager or poor manager for the P eleven through a P thirty three so that is Sir the total gain of eight hours per week.
Which is it was very good performance because this was fairly difficult to work for all of the managers involved.
Now moving on to our HAI roadmap and our plans for the future.
Where are the using the tool it is in production in our patent department we also have some projects or.
Our automatic classification of trademarks and industrial designs is also a project to a preclassification of patent applications will use the same tool.
And we have a chap boat on our side, the NPS sides are quite like the one that the Australians have.
And we also use WIPO translates and IP cat.
We have an IP cat pilot contests on the way at the IMPI obviously we are ready to work with any other officers interested parties.
So many thanks were listening to me.
I am Jonathan wit is my email if you have any requests on Aye tools at I am here you can write to olivia and he so you see his email address up here on the slide many thanks.
I would like to invite the Delegation of Spain.
Thank you.
Thank you,
Good morning, everyone.
The same I should like to thank the Secretariat for giving me this opportunity to make this presentation and.
I would like to congratulate Japan and I did see kingdom and France who have given us much sure.
Material for thoughts in their presentations my name's lapelovna on the head of the patent examination unit Sir of mechanics and construction in the Spanish offers of patents and marks.
I have done added presentations on other topics you may know me already with regard to artificial intelligence today as you are aware this is a very hot topic for the last two.
Yes there's been a conference on artificial intelligence and patents in one or other place in the world for the last eighteen months said the department of our office had so it.
That we should investigate what was available in terms of artificial intelligence for our department and find the areas of interest that could benefit science and the methodology in our office.
So we put together a group in our office with representatives from all these sectors of patents marks.
IT rights the law legislation et cetera our objective was to compile all available information on artificial intelligence and at say to what extent it would be applied in our office.
So we found two areas of interest which are the
Items that we would have two exchanges and experience on and at a later stage next this summer will be discussing this on the basis of a joint proposal by Spain and of France.
At previous session of the committee the vest of the areas of interest that I want to share with you to day.
It gives you an I dare the later presentation I will make is called patentability within patentability we have we have to distinguish the following sub sets of relevant.
Arias patentable subject matter what happens autonomously using artificial intelligence who has the right to these inventions afterwards.
I have generated inventions and this is a very topical issue because as a result of a patent application.
Where the inventor is in algorium using artificial intelligence what happens another relevant topic is what happens with prior art that is generated by artificial intelligence there are Webb.
Pages that are automatically generating in an ongoing manner, the prior art.
And this has to be taken into consideration, provided that sufficient disclosure and in terms of assessment of inventive step who is the.
Persons skilled in the art is it Aye is a day person.
They have access to these tools in terms of sufficiency of disclosure, this is also a very sensitive topic when it comes to inventions generated by a I.
Because of a lack of reproducability of what happens between the entry the data and the art put this topic cannot be discussed more using Brazil's proposal for example.
And finally this is a topic which is certainly more removed which is liability in patent infringement by a I.
What happens if an algority infringes a patent law who is liable.
You can talk to electronic pets, you can talk about electronic personnel, it is relevant by the European Parliament, compulsory
Insurance that might be relevant to and another topic that was pinpointed by this group was the use of intervention at its intelligence for patent examination.
And that is why we are going to focus on now we know it says.
Back to there are areas in which you can use the artificial intelligence in patent offices in the first of them is patent document classification, not just for patents, of course, you could also classify marks as well and go.
Other patentable material also the development of virtual assistance so called chat buds for Internet searches that go on to the website.
ODOMATIC a translation of patent texts.
This cannot be used in formalities, examination and exclusions from patentability.
And prior art such this is the most stem impressive area where I I can make huge progress and I notice because that for many years I was a patent examiner.
And finally you can use AI for so called writing assistance to help you draft a letter which usually.
Is carried out during the process of granting patents now going to focus on some of these, the first is patent document classification, and
The Delegate of France has already given us in our view.
Of this matter any of us will encountered the problem that to properly distribute the patent documents you have to preclassify so.
That it goes to the most appropriate group of patent examiners what we did instead of developing our own agorathem we thought it might be more efficient to use.
And algres them or a program based on an i.
Which share WIPO developed it is called IPC cat and is not as as when I have had the potential arrives in our IT system file management system.
Ultimately we extract a summary of the application and here you have an example.
This is a summary of an invention on a trap for insects and automatically this.
Program A I program which also works in Spanish it provides us with three possible.
Classifications or categories according to the international patent classification system then we have subclasses and it offers us the most appropriate which is.
The one finishing in aim which is appropriate for insect traps so based on this program using this program the patent application can be.
Allotted to the appropriate group of patent examiners.
As the French delegate already said thanks to these algretoms we can hugely reduce the number of hours needed to decide which patent examiner should be examining the file.
Another area in which artificial intelligence is facilitating the work of our patent offices is that of automatic translation of documents.
The Patent Offices have limited knowledge of languages in our office for example all that decaminazak conversing in English some also in French very few of them and no German.
Some of them might know Russian or China but many may not and thanks to these tools.
You have the two instruments when developed by the EP, WIPO Translate, produced by WIPO.
Thanks to these two tools Sir we can have access to para documents which previously we could not access or we could only do so via own.
Summary documents in English and an advantage of these tools is their specialized in patents.
One of the problems that A I raises as you can not always access available data in a free manner and in the case of patents there are a lot of documents in the public domain.
Which are freely accessible and enable these translators to then continually perfect their their craft we also have another project.
Which is to translate all of the patent applications that come into our office that have to come in in Spanish and we have to translate them into English and why well.
Because the best such engines work in English so he is an example for you with regard to a bio.
Climatic Co construct it automatically translated into English a biochomatic building.
And this never has to have the key words that once we need the relevant words and we then use this text in an i search engine said.
As I explained before and one of the problems that to these translators cause is this data is confidential.
So we have to install these translators in our own on our own service.
And now the most amazing topic when it comes to additional intelligence.
Is that of prior art search what they these tools can do so how do we look for prior art in 19.
Seventy in the United States office at the time they printed the patent application documents and they kept to them in accordance with their classification and they all went into.
Boxes did some of them are electric motors and they were all kept and saved those related.
To but Lebanas were kept in another area and the patent examiners had to go to the boxes and register all of the relevant patent documents and now.
And this is how it used to work and when I excited the Spanish patent office in the 1990s, people were still working like that for some of the Spanish documents.
And then the search engines appeared that were in IT systems with the patent summaries and
Now we have arrived to the age of artificial intelligence so how exactly does it work usually when prior art such is carried out.
And you will say this at the bottom of the slide, you will see a white section where we copy you copy the full text of the patent application.
Just the claims but of the claims you can use key words as well to substantiate them.
And sometimes the search engine that used to be semantic such engines give us a list of most relevant documents in order of relevance as you see on the lower part of the page.
And when this technology appeared for us well at the beginning there were a lot of expectations because as I said.
When new technology appears it seems like magic and it seemed like it will solve all of humanity's problems and so expectations were high and the idea of searching for prior art using an eye was amazing on weaseward.
That It was going to replace the patent examiners even perhaps.
And this reason we started trailing it and we had come up with commercial firstly such engines based on an i we tried eight of them eight search engines.
And the results I must say was a little disappointing, but because perhaps our expectations was so high.
To try out these AI such engines we measured two types of variables firstly we measured the recall function.
This means we used a search engine a patent examiner a human patent examiner did it and we saw the daughter appearing.
We used the patent examiner, we compared these search times found using an iceage and gene.
Sometimes they were not matches or there was one match, one would be 0.1 in terms of recall.
Et cetera I also measured what we call the precision precision means how many of these documents that are recovered by the algeridom or the search engine the A I so change in many of these documents.
I really patter the prior art when it comes to the invention in question we also solve.
That these searches work much better in the cases in which we are, they do not use drawings or pictures.
And another step that we are currently working on is searches using drawings or, in the area of marks, it is already and
And like a logo and based on this we can look for other luggers you can scan the drawing.
And that is provided by prior art related to these drawings this was still in a very early phase with regard to this technology so let me tell you about our conclusions based on these experts.
The moment it is not possible to rely on AIbased patent search engines to find the relevant prior art.
However it is true that these such engines sometimes do provide a good starting point particularly fun.
Offices where the patent examiner is not very specialized because it gives them some relevant documents that do allow them to then deepen their knowledge and be more efficient when they carry out.
They are search and reduce the ultimate time spent on the case, however, we are still far from there's algreds replacing patent examiners.
But we will get this we will get there eventually and probably.
One day and algorium will replace the patent examiners, these are some of the famous of them, I understand, Damjan, you have the
And that is in the future we will also have one or two of these amazing agrims that will replace patent examiners.
The imports of this technology is that it will enable us to address the constant increase of that duplications that we are witnessing.
And this is also affected by the use of A I in a generation of these inventions as well so thanks to an i.
We are going to be able to deal with an increasing number of patent applications that are based on AI inventions themselves.
So this is my presentation of what we are doing in our office of patents and marks in Spain with regard to AI, I am looking forward to hearing
The other presentations on this topic and we are very happy to work with other offices and you have any questions or any comments, I would be very happy to hear it, thank you very much.
Thank you, now we have a presentation from my colleague from WIPO.
Thank you,
Thank you,
So everybody, my name is Bruno Prita, I am heading the advanced technology application center.
The economies attack in WIPO where we build WIPO translate a high-based machine transition tool and I am going to present you, WIPO translate and other.
AIrelated tools that we developed here in WIPO.
So what is artificial intelligence usually when we say artificial intelligence we mean deep supervised machine learning.
So and we are doing deep supervised machine learning, so much in learning means we learn that the machine is learning from examples
Supervise mean that those examples are known from human generated data for machine translation we learn from previous translations that were high quality.
Done by a human beings and deep her because we use multilayer artificial narrow networks this new technology in your networks are lowers.
To learn from a lot of data and to try to guess what is the most probable category what is the most probable translation for this new new example.
So what do we do in WIPO, roughly we have got three pillars, we do text processing, we do image processing and speech processing.
In text we build WIPO translates new role machine translation, we are also doing classification of text and also
A experiment menting with transliteration and we are also experimented with chadbot.
About image we do image classification we look for trademark image and we try to classify automatically.
This image into various classes could be US class of Vienna class and this allows us also to do.
EMAGE search we can provide an image unlooked for similar images images that have a similar classification.
And about speech we do speech processing like the speech two texts.
And you are lucky because this week we are running this experiment for the SCP where we run this speech to tax on, I hope I will have some time to demore how it works.
So speech would like to allow you to search in videos.
But there are a lot of other use cases like for hearing in purpose and, for example, also, and we can even explain and we speech to transside the text.
I mean somebody who talks in English and then we all put the tax in Chinese for example.
So a strong disclaimer that we try to highlight each time we present machine on in tools or AI, so any machine transition or speech processing,
Including the one we develop.
Is not designed to replace the need for human interventions so that the human still needs to be in the loop, so this is just a tool to aid human in making the intervention more effective and efficient.
And you will see if you explain and with speech to text it is not accurate it is not a humanly produced text.
But however it has a lot of advantages its extremely quick at three me cheap for example.
So WIPO translates, we started WIPO translate in 2011, so we have got quite a long experience.
In dealing with machine translation so previously well it was using another technique but since 2016 we are now using newer networks.
And for information we use newer Networks only five days after Google published the Nero Network where Google transrate.
So we are only a doctor of this technology, it is available on patent scope.
And it provides instance we hope quality I mean it is not human quality is not publishable quality but we hope it has a certain quality in ten languages.
So best would be to do a demo let me try to switch.
To a demo so if I go to patterscope.
Our search engine in WIPO something you may not know but you have got to cross in all expansion which allows you to search in a language and retrieve patents in the different language.
So, for example, I search in French as well, raise a place as an electric raises.
The electric shaver with radio says searching French I submit the query.
This is going through the crossing all information which we will tool and it arose me to retrieve patents in different languages including here for example in Chinese so if I click on the first one.
So suppose you are like me you do not read Chinese.
Then you look at this, this patent is all in Chinese and then you can activate WIPO translate, and here we choose to translate this patent into English.
And then you see it is an utility model that wireless power supply shaver and you can look at also the description the description is all in Chinese,
And you will see WIPO translate in action and then he translate all the description into English, so it is not under a person accurate.
But you see the advantage is extremely fast and it provides you an access to a gist translation something that allows you to understand what this patent is about.
So this is WIPO translates, you can also hear you can also translate your own abstract from English into
Into Chinese or the other way run from Portuguese into English, we have got all the ten official languages of the PCT in this.
Sorry. Yeah.
So WIPO just later on patent scope is quite used, in fact, we translate 2.5 million words every day.
To give you an idea the United Nations translators.
They translate 400,000 words per day so it is quite a huge amount of words that are automatically translated so we can.
I think that is a good sign that it is useful for the users.
And if you look at the language pears that are used on patent scope it is quite interesting, the most important one is Chinese into English and Japanese into English, namely English and to Russian.
English into Chinese, so it is not only from
None English into English is also from English into non English so it is all it is used by Chinese speakers to understand English patents also.
And something to note is 65% of all the translation from Asian, which is Chinese, Japanese and Korean.
About that was about WIPO translates, we also do IPC cat, it was mentioned.
By the French by INP and by Spain, IPC cat we.
We look at existing classification from title and uprights and we try to guess what is the best IPC code for a new abstract.
So it can be useful automating automatically roots the patent documents but also other technical documents.
It can be used to improve the consistency of classification.
And it can be used to do IPC reclassification or IPC classification when no classification is available for scientific articles, for example.
And this is available on a public Web into face it is based also on newer Networks.
I will not go deep in to image similarity search but
It is also something we develop in-house in WIPO where we learn classification of logos of trademarks, images.
And then for a new trademark we are able to guess what is the most probable classification for this image.
And here what you see is an example of search on the global brand database in WIPO you can search for you can provide an image like.
This image you see here and then it retrieves you similar trademark images.
And you see it is extremely accurate in that example and it is better than the previous searches that were based only on shape or corals.
Here it is really it is a deep classification tool and you can feel its definitely more curate and it allows you to retrieve complete different images.
But that still contains leave.
We also experimented with chat boards in WIPO.
This idea was too for the world was to help the WIPO service dusts, and we collected the estalrecory males, and we created the question answer
And then we try to create an AI tool which is like a translation tool.
Instead of providing English and for a language, we provide question and answer and we try our tool on that.
To be frank it was not a great success I mean we had some results but some of them were not good, so here I give you
The first example, the question we do not see nor manage I were, I and in our Madrid portfolio manager, whole.
To add our brand to your account on the answer was it should be okay now so that is a completely wrong answer from the A I base tool so it is quite funny answer but not really useful.
But you can see that sometimes it provides very good answer like where can I find place WIPO in Morocco.
And you can see that the English question is not exactly good English, where can I find place WIPO in Morocco and you see that the answer is quite accurate.
The world Intellectual Property does not have an office in Morocco and et cetera, so it is really balanced so it would need further investigation.
Now the speech to text is not related at all to patents, but it is related to this conference because it is going through to this exercise.
So and you can access yourself in WIPO and/as duty like speech two texts and I assure you a demo.
It is quite risky because it is.
It is really a live demo so if I go to as to tea so it is a public interface that you you can use yourself and you see.
That we have currently two sessions of the SCP 31 it was yesterday morning and yesterday afternoon, so let's try yesterday afternoon, so I cake on that.
And then you can see yourself in the conference and you have a sliding window which allows you to see the speaker.
The audio, if it is a general statement from regional coordinators on quality patents.
Itself with statements made by Croatia on behalf of the CEBS group and by Finland on behalf of the EU and its Member States.
So you have to understand that this is yet another machine learning tool.
We took all the WIPO conferences that took place and we took the Ojo, we took the verbatim that was produced by human.
And we learn from that, so when now we have a new speech, this machine learning too is able to produce the most probable English text.
You will see that it is not and it was inaccurate forgivers it will never be 100% accurate but the good news is available.
One to our after the meeting, so it is extremely quick, it is available on the web, you can use it, and on top of that,
The fact that we produce a text from video it allows you also to search the text and then suddenly those conferences become searchable.
So if you look for it.
GRULAC, for example, then you see all the statements that were said by you, Delegates or from the Chair, so
So Mexico has the honor to speak on the other the group of countries, no, no, let's say.
I take this and then I click and you would like to say that we that can assure you of the support of GRULAC throughout this week and.
We are so and you can search text, you can search the text or unfind the corresponding video or audio.
There is even one step more, we can access automatic translation of the produced text, so
Please before reading it have in mind that it is this is an automatic transition of an automatic transcript.
So it is yet another level of quality, it will not be 100% accurate far from that, but it provides you an access
Two a speech in English in for example French or Chinese or Arabic so you have the same here you have got the French and for example a dotterdomen year and a phometique.
I had in other areas associated with the IT and we do know that product life cycles in these industries are particularly short.
So if the administrative lice cycle is longer than the technology life cycle.
So when you click then you access the English video, but you have a way to get it in French, and on top of that you can
Search now in different languages you can search in French for example you can what can I search for?
So this is a French word and I search in French and then I see all
ESurgy to refine the same group as denuable technology if I click on that then I will access the English flow as because the mining of patent fire to face on new technologies.
As a lounder is between Tehmore you can see some some arrows like Piton 5 was badly translated at 5 Patents, so that is
Once more we have to accept the low level of quality, it is not publishable quality from that, but it allows your rapid access to the information.
So I invite you to look at the speech to text and in fewer you will be able to see this conference in a speech.
So an eye on patent what are ongoing, what are the future work in WIPO, so we want to extend WIPO translate to other languages.
And to other language bears currently which on set everything into English English into everything maybe in the future we might translate directly from Korean into Japanese for example or for all from Spanish into French.
We want to extend it to written opinion and international search reports we are currently offering WIPO to say for that, but the models are not optimised, but we will optimize it for that.
We are also adding a new functionality in patent scope that Markush search he will be able, we hope, to search directly Markush
Temical formulae, and we also experiment to search by tax similarity.
But this is purely exploratory, look, do some word on bedding, sometimes some money, and look for similar sentences automatically.
Which very man with the name consideration also name cleaning on maybe name proper names standardization.
A quit slide on our strategy in an eye or machine earning.
We came at what makes us such successful projects is we should carefully choose the potential successful projects it is not that we can apply an eye to everything.
Especially when it is machine learning it depends on the data you have if you learn on few data you are not going to learn anything.
On top of that we should accept the risk of failure and accept that the expected quality.
Will not be human quality most of the time even if sometimes it could be more than human quality for some specific use case.
It should be business-driven and it should be well integrated which is a difficult part of.
Of this kind of product because there is a change management we should not be neglected.
We, for example in WIPO, we aim at quick prototyping, when we have an idea, we try to have a quick prototype in a window of six months, one year,
And then looking at the prototype, we can decide, we continue this project, we stop it, so we have the right to fail, but we want to fail fast
Not to go into the wrong direction for years.
And faith us to innovate faster. We are doing survey of technology, we have an academic network in WIPO with chosen to use only open sources.
Because in this kind of technology usually, usually open sources are the most active community and other closer to the state of the art.
And on top of that we should gather data we should fill the data we should clean data so you have understood.
Everything relies on the data its machine learning so of course the data is the golden mine of machinery.
So AI what WIPO can assist IPOs.
So we think that maybe the tools we developed are valuable externally.
It is in the IP industry, the scientific community, but also the Member States and IP offices.
We, for example, install WIPO translate in the Korean patent office and this is a kind of service that WIPO can provide to help IP offices.
We also provide our tool to UN family.
WIPO translates in other international organizations and also speech to text is beginning to have a lot of interest from various other United Nations
Organization but also other international institutions, so we WIPO is trying also to
To fight, let's say, the global digital divide is then committed to provide inclusive access to where WIPO AItools.
So we are currently following a demand-driven approach in licensing of AI tool, so if ever you are entrusted, please do not hesitate to contact us.
So we continue to drive this cooperation by organizing conferences and workshops.
And we want also to share our AItools and, of course, sharing data is also part of the
Of the game because we learn Chinese to English, WIPO translate using the Chinese data provided by the Chinese Patent Office, for example,
Thank you very much and thank you to all of the presenters this morning.
But some really interesting presentations then a lot we can all learn from each other I think so what I will do now is ask if any one has any interventions from the floor on this topic.
Well, Czech Republic, you have the floor.
Thank you, Madame Chair, the Czech Delegation wishes to share the information on the use of artificial intelligence.
For examination of patent applications as well as for improving services provided by the Czech IP office.
First, let me mention that the Czech Republic is fully aware of the high importance of artificial intelligence.
It is used for the future development and competitiveness of national European and global economies and societies.
Therefore the national AI strategy was approved in March 2009.
This strategy specifies a framework of priority objectives and tools to support AI development in the academic public and private sectors.
Mutual cooperation and international engagement the national AI strategy follows and supports the innovation strategy of the Czech Republic.
2019 to 2013 and its mainly linked to its part dedicated to digital Czech Republic production and services.
The innovation strategy was approved by the government in February, 2019, this is the first time that the policy instrument.
It includes the individual pillar dedicated to the protection of IP based on the strategy, the Czech IP office is currently working on the IP protection concept which
Proposed measures will be elaborated strategically and in more details to properly implement goals of the IP pillar of the innovation strategy.
But back to our main mission today, also the Czech IP office recognizes the high potential of artificial intelligence.
And its use in IP management as well as in enhancing services provided by the IP system users.
By inspiring with other IP offices and based on our needs we are gradually increasing the use of artificial intelligence in different activities of our office.
I will start with in initiatives improving the IPR management.
Starting from 2021 we plan to launch internal AI examination support which will help our examiners with the file distribution.
Classification of patent applications and Braille art search.
Next group of activities is more focused on improving our public services first the office is currently working on creation of automated IP whole desk.
Which will be non stop available to our users the idea is to start with the provision of IP related advice in general.
Later we wish to enlarge the service and provide also the information dedicated to procedures of different IP applications.
In this context we would like to take advantage of the cooperation with the Czech technical university in Prague and introduce Chedboat to improve our health desk service.
For the purpose of enhancing the service even more we have also analyzed the use of voice recognition.
But the reality had shown that it is still rather problematic for the check language.
In terms of the introduction of automated search and classification system, we have run the proof of concept.
It confirmed that such a project is how it is helpful.
And really needed the testing system is already available on our website and offers the analysis of invention to potential patent users before filing.
And suggest the classification of that invention.
The preparatory phase of this project will take place in 2020 and then we plan to launch the service in 2022.
Finally we are currently considering the implementation of the machine translation project.
It will offer translation of all outputs of our national IP databases and make their content available to potential foreign users with this I thank you.
Thank you, next I have Singapore, you have the floor.
Thank you, Chair.
As what are in the report, WIPO technology trends of the intelligence.
E I related inventions are booming and shifting away from shifting from theory to commercial application there is increasing demand for your technologies as and the prizes explore new ways to apply your solutions to keep in a competitive edge.
For public agencies by the IP office the technologies offer new capabilities to better serve as stakeholders.
The Singapore delegation would like to take this chance to share that Singapore recently launched the national AIR strategy just last month.
It is a key step in our smutnation journey the strategy paper is spelt out the Singapore's plan to deepen our use of air technology is a transsum of economy.
Going beyond just adopting technology to fundamentally rethinking business models and making deep changes to prolivity gains and create new areas of growth.
The Singapore IP office has taken steps to leverage here technologies to better serve our stakeholders we launched.
Of mobile application for trademark registration called IPOSCO in August this year the application integrates air technology to power such as for similar mucks.
On the I pause register and transform the way applicants perform detrima falling which is now possible through the more body vices and the falling can be done in under ten minutes.
The team that developed the apopostal learned that in meeting the needs of present and future innovators he was crucial to not just be able to leave it in technology but to also adopt a human century.
DESY approach to create a seamless experience for the user.
Our office is also developing an eye for patent classification and prior art search to improve our efficiency to in turn benefit of the holders.
The development is a collaboration with the Singapore aepogram office caught the IASG and experts from a research institution in Singapore.
We have particularly involved our examined patent examiners in the project to lead the project to design project development phases and also learn the skills to improve the air model they will come up on the project.
We have embarked on this project a few months ago and we look forward to receiving some results in the next twelve months.
We believe that AI has huge potential in transforming the way IP officers operate and service that we deliver to applicants.
We thank the IP offices of WIPO for the earlier sharing and look forward to information, experience and exploring possible areas of collaboration among Member States in this SCP and in future sessions.
Thank you, Madam Chair.
I have a long list of Delegations waiting to take the floor so if we take a short coffee break and then we will come back for the rest of the interventions think.
Thank you,
Thank you,
Thank you,
Thank you.
Thank you,
Thank you,
I welcome back everyone.
I welcome back everyone, so I have a list of interventions continuing our discussion.
On my date, I am so continuing our discussion on the use of artificial intelligence for the examination of patent applications.
And so I have Germany when the USA Iran Indonesia Australia and Chile so Germany you have the floor.
Thank you, Madam Chair. As this is the first time that we take the floor on behalf of the German delegation, I would like to thank you and the Vice Chairs for leading this Committee again. We look forward to a fruitful meeting under your guidance.
We also thank the Secretariat for organizing this meeting as well as for the meeting documents which are once more of the highest standard.
We were excited to follow the excellent and very informative presentations by Japan, United Kingdom, France, Spain and WIPO and we appreciate the opportunity to provide information.
on the application of artificial intelligence in the patent administration and patent examination at the German patent trademark office, GPMA.
The DPMA is currently developing two AI applications for the use in the patent administration and patent examination process.
Firstly a tool for automated patent classification according to the international patent classification IPC.
Has been developed. The new classifier at the DPMA provides an automated pre-classification of incoming patent applications.
It will improve the distribution of patent applications among the examiners within our electronic patent application system.
Furthermore an interactive classification service assists the patent examiners by providing three predictions each with a numerical confidence level for text input.
The new classifier also is or is planned to be applicable applicable for re-classification (the support for the introduction of new versions of the IPC),
the maintenance of patent literature and further applications like IPC at classification of patent documents of foreign patent offices that do not classify according to the IPC.
The tool provides a classification depth up to the IPC subgroup level. The training collection is balanced across the IPC hierarchy.
The more documents for path of a IPC tree are available, the deeper access to the nodes of the tree structure can be achieved.
Until today the system has been trained with publications of patent applications granted patents and utility models. The supported languages are German and English.
The German model has been trained with about 840,000 DE and EP patent documents. For training of the English model we used 860,000 EP patent documents.
The classifier is implemented as a web service. It has been integrated into the digitization center of our electronic patent workflow system.
After a successful trial phase, it has been set into production for the classification of patent and utility model applications and for interactive classification on 1st October of this year.
Secondly, a new patent search system is being developed that includes a cognitive search. The application is currently being evaluated by a selected group of examiners of the DPMA.
The cognitive search client provides a prior art search or pre-search function which automatically a generates list of content related patents for a given document or text input.
Hereby, the cognitive search client allows searching, (i) in a simple search mode based on key phrases; or (ii) in an advanced search mode, based on one or more entire documents and/or document parts like selected paragraphs.
In the documents of the result list words with similar meanings as the input text are highlighted. The system has been trained with publications of patent applications granted patents and utility models.
In order to allow for a cross-lingual search, two models have been prepared - a German model trained with 2.6 million DE and EP patent documents and an English model trained with about 14 million US and EP patents.
The text fields title, abstract, claims, description and specific texts provided by the DPMA examiners have been used for training.
In March 2019 several functionalities have been improved and added in a new version of the cognitive search, such as the automatic synonym determination,
the cross-language search in German and English, the search in text parts and in entire documents, the storage of the search, as well as interaction possibilities such as adding additional search terms or key phrases.
As mention however the patent search system is still in its evaluation face.
Further work on the project aims at training the classifier and the cognitive search with the whole document set of the DPMA that consists of about 118 million patent documents.
Particular future objectives are for example the improvement of the cross lingual search like merging the results produced in different languages into one result list,
integrating non-patent literature at model training as well as the recognition of chemical entities. For the future the DPMA considers that in addition to the classification of applications and prior art search,
AI could be of particular relevance in the fields of (i) translation of patent documents; (ii) image search and recognition, and (iii) search in chemical formulas and sequence listings, thank you, Madam Chair.
Thank you, next I have the USA, you have the floor.
Thank you, Madam Chair, I want to thank the offices who presented this morning, it was very interesting, very informative.
I am I was struck by as we get into what I am going to say that the US PTO is up to basically the duplication of effort that is going on.
Ah we all seem to be working on the same kind of things and I commend WIPO for all of their work and their willingness to share with the other children the stuff that they have done.
Our office is actively investigating the use of Aye to help develop solutions that it can improve examination of patent and trademark applications at our office.
In fact we are looking to higher a senior level artificial intelligence expert.
Ah this person's rolls and responsibilities will include providing technical expertise in an i.
Providing input on high level decision making regarding production and operational implementation of A I in our infrastructure and architecture.
They will advise on automation strategy and identify ways to leverage AI in our business processes and they will develop an AI roadmap for the US PTO.
Ah AT this point our office has three primary goals with respect artificial intelligence to improve the examiner's ability to obtain the best prior art.
To create efficiencies throughout the patent prosecution process and to optimize development and delivery of information technology tools.
I am the areas that we are currently focusing on and again these are somewhat duplicative of what the other offices have shared.
Ah we are developing a trademark image search tool for both image and design code searching we are investigating the feasibility of industrial design image searching tools.
We are evaluating the use of A i in presearch tools that can help examiners understand the prior art landscape with respect to a particular application and identify potential areas of search.
And finally we are developing an auto classification tool to assist with the routing of applications.
The USPTO is very interested in learning more about what the other IPOs are doing with respect to AI and having further discussions on the topic and related issues thank you.
Thank you, next behalf of Iran, you have the floor.
Thank you, Madam Chair we are appreciative to the speakers for sharing their experiences on the use of AI in patent examination process in particular interesting presentation by colleague from WIPO Secretariat.
It is a good idea to share the experiences and views among Member States on the benefit of using AI to increase sufficiency of the patent examination process aiming at advance the quality of patent.
AIas one of the main aspects of the new technology development has a great potential to affect the function of IP offices in general and patent examination, patent application examination in particular.
Meanwhile, it is a fact that these new technologies in advanced level are available in a limited number of countries, this is an important element which requires due consideration by the committee.
We believe that the capacity of the offices to use AIC challenges of the LDCs and developing offices and the method of delivering technical assistance to use.
To these offices to enable them to use AI should be discussed in future by the committee, the global technological gap and the importance of narrowing the current gap necessitates cautious approach by the committee in this regard.
We also need to consider whether using AI will reduce the cost of the patent examination process or it will increase it, potential raising in the application fee has a lot to say here.
The Committee needs to explore whether the use of AI could be extended to all aspects of the patent grant process or it would be limited to particular aspects of the process.
The territorial application of the patentability criteria which identifies different threshold for instance on knowledge criteria.
Is another element which affects the role of the AIbased examination process.
Apart from the consequences of using AI on human resources which requires serious consideration by other relevant fora we are of the view that substantive examinations should maintain.
A human interface based on the review by the patent examiner. Thank you, Madam Chair. Next I have Indonesia, you have the floor.
Madam Chair, first of all, we would like to thank all of the speakers for sharing with us their experience in integrating the patent application process.
Admittedly in the case of Indonesia we are still now infancy in integrating AIin to our patent application process but we are very much interested in technology and the opportunities it presents to improve efficiency as many of the speakers have stated.
Nevertheless, as swiftly application of any new technologies, it is always better to err on the side of caution, due consideration to sequencing in pacing must be given to minimum possible errors that are to be expected in all for generations technology.
In this regard, we fully agree with the speakers in which he was locally stated on the need of keeping human in the group, we fully support the principal evolution rather than for revolution.
As it has been pointed out that the accuracy of the idea is still lower than human dispatch admittedly the system is improving rapidly.
But we must bear in mind that in promoting efficiency must not forego effectiveness, this is in line with what we have discussed today in which a quality of patent is as important as the patent process itself.
Having said that, Indonesia is very much interested in collaborating with countries that have integrated in to the patent application process and we should more and sequence corporate the system taking into account different levels of development, of course.
Furthermore, as we move forward with this technology, if any of the speakers could share with us the cost associated with setting up the AIsystem, what is the expected budget allocation and how much cost efficient is the system compared to the old system.
Lastly, we are very much interesting with the presentation made by our same colleague from the WIPO Secretariat, particularly with regard to the text translation technology.
As many of the patent laws and publication are written in the respective members' national language, would it be possible for the WIPO Secretariat to extend the use of its translation technology to the transmission of those laws and publications?
Of course, based on one request as well as open sources, this would under the B be of a terminate benefits to the members not only in terms of efficiency but also transparency, thank you, Madam Chair.
Thank you, next I have Australia, you have the floor.
Thank you, Madam Chair and good afternoon colleagues, first of all, we would like to thank Japan, UK, France and Spain for their respective presentations.
Of the work they are doing in the field of AI and thanks to WIPO as well for their presentation as well as the life demo of the tools in the AIspace. That was a very interesting work indeed. Now, IP Australia, like a number of other offices,
Are in the early phases of adopting a power tools to assist examiners and we are very pleased to have the opportunity to share IP Australia's experiences with the eye and machine learning particularly noting the impacts of these technologies on.
IP rights administration and examination now as we all appreciate the road to fully adopting an eye into our administrative systems is a long one and there are crucial learnings along the way.
We would like to use this opportunity to give you all a brief update of some of the current initiatives that we are currently working on, the first one is a patent or classification tool.
We are working on a system that automatically look allocates patent applications to the current examination sections across hundreds of technology groups now Pandallum.
The Patent or the classification uses machine language, machine learning technologies to analyze the contents of each patent case in unstructured VDF documents and predict the relevant technology groups and hence the appropriate patent examination section,
As this classification is at a very high level, upon receiving the application and reading the description, the examiner checks that the application has been allocated to the current.
Technology area and actually examiners can manually reassign to another section if the initial classification is deemed not to be correct or accurate.
The other tool that we are working on is called an ultimate preliminary search tool that searches the potential citations again saving staff, staff time, examination time.
The automated preliminary search will carry out an automated search of published patent specifications by using search queries containing the names of applicants and inventors and IPC and CPC symbols.
And the tool was just released in November and training of examiners is actually currently going on in IP Australia.
The third tool that we are currently using is called the patent family member analysis tool for examiners that is currently.
Under development aims to retrieve published patent member and family member information of a patent application and present it in a way that allows for an easy visual comparison of claim sets.
So the last 8, 12, 18 months, IP Australia has been implementing AItools as well as making changes to legislation that allow us to use these tools in decision making for IP administration.
Finally I would like to finish by looking at changes we made to a legislation that allowed for computerized decision making.
I Australia passed an amendment to the patent act to allow the commissioner of patents to arrange for the use of a computer program to make decisions.
It exercise powers and comply with the obligations under the legislation and currently the administration of the patents act and regulations require the commissioner to make a large number of decisions.
Some of which are straightforward others are more complex the use of computerized and computer assisted decision making will improve the timeliness and accuracy of decision making and enable staff to deal with more complicated matters.
An automated decision making governance framework has been developed to establish the structures and controls required to manage the automation in IP rights administration.
The framework applies to any decision power or obligation under the legislation that is to be automated by a computer program.
The internal governance framework is important as it ensures that the commission is power reduced judiciously and provides a level of comfort to our customers.
As you can all see these initiatives like other offices are focused on efficiency against which hopefully would lead to be the timeliness.
I am something that Dr Gurry WIPO was Director General emphasized in his opening remarks yesterday.
To summarize, all our initiatives would like to say that we are keen to harness the benefits FAI and automation in order to achieve efficiency against all aspects of examination and administration.
To improve outcomes for our staff and our customers thank you thank you next I have chilly you have the floor.
It is just in it thank you very much Madam Chair since the first time that we are taking the floor here we would like to.
I will tell you that we are very pleased to see you chair in this Committee and we hope that you will have much success in leading this meeting our Delegation would like to thank the Secretariat for the organization of the information session, the use of
Artificial intelligence and the examination of patent applications to the information such as neighbours to see and understand how the.
IP offices have used artificial intelligence in patent applications.
In an innovative manner which enables us to include better efficiency in the processing of patent applications as well as to help the patent applicant and.
The inventors for Chile the implementation and use of interartificial intelligence is it extremely.
Relevant topics so as we have said in our ministries, science, technology, knowledge and innovation, and they were drawn up aid.
Work plan on artificial challenges which envisaging the initiating of a policy and plan of action for the year 2020.
Furthermore our National Institute of industrial property E Napi has already using artificial intelligence belonging to the carry or PEARLING.
In the procedures of marks through a search sentence to chambers to use the comparison of images of trademarks as including the similar tools for the remainder of the services it provides.
And this includes in the procedure for the processing of these application patent applications this will be included in Napis also working with other.
IP offices on this topic, exchanging information and participating in informal groups with countries of the OECD, and it is because of all this, an exchange of information which has been generated in this
Point, it is very useful if it says the rule of artificial intelligence for patents in our office. Thank you very much. Thank you, next I have Pakistan, you have the floor.
Thank you, Madame Chair, we deeply appreciate the manner in which you are conducting the session and we also thank the presenters for sharing their experiences on use of AI tools for patent examination process.
The use of artificial intelligence for sporting examination of patent application as wider amiss brought the technology remains ahead from the expertise level of the skill set of IP offices in many Member States.
The examination of patent applications is a very complex process, the judgment of patentability varies from one field to another field of technology.
Especially the claim claims language of invention related to chemical and pharmaceutical products where multiple variable combinations are claimed such as Mercosur claims.
Another example may be of very long gene sequences.
The analysis of such inventions become more effective with AIbased sport tools, we appreciate WIPO for offering ABS tools for patent classification, image search for trademark and language translation tools.
There is still need to take further steps in guiding Member States in devising both short-term and non-term A policies for national IP offices.
To integrate AIbased tools efficiently we suggest the Secretariat to enhance role of WIPO advance.
The technology application center to assist Member States in this regard, we also suggest WIPO to integrate AIB solutions in industrial property automation system that is offered by WIPO to Member States.
We encourage Member States to continue sharing national experiences regarding implementation of AIB solutions. I thank you, Madam Chair. Next I have Brazil and then India. Brazil, you have the floor.
Thank you very much, Madam Chair, we would like to thank the Patent Offices of Japan, the UK, France, and Spain and WIPO for their interesting presentations.
As other patent offices Brazil is beginning to develop AA tools in order to improve the automatic search for patent prior art.
Currently a partnership is in place for the creation of a search tool that uses similarity and it is based on the use of key words and classification of documents.
With the aim to assist patent examiners so as to expedite the examination of patent applications the Brazilian patent office has already introduced narrow networks in the process of its management system.
In order to allow the preclassification of patent applications and subsequent distribution to the technical divisions through an inhouse developed programme.
Brazil is very interested in cooperating with other patent offices with a view of improving our examination tools on artificial intelligence. Thank you. Thank you, next I am India, you have the floor.
The Delegation of India reaffirms the views as expressed in the earlier SCP with respect to the revised proposal as submitted by the Delegation of France and Spain to SCP 31/9.
The Delegation of India also expresses sincere thanks to the Secretariat for organizing the sharing session on artificial intelligence.
It is no doubt that the progress in the digital world is in the near future would be based on the pillars of artificial intelligence, block chains, et cetera.
And accordingly the IP world cannot remain a mute spectator to the inventing technological changes as being impediment under the full industrial revolution.
Hence the Delegation hopes that the setting that the sharing sessions in this regard helps the Member States to carry it back to their gap to the foster lives.
Of artificial intelligence for IP the Delegation also believes that this sharing session will be also helpful to developing countries and the IP policymakers as a whole, however, WIPO should also take into account
The fact that the fruits of this new technology should not be limited to mere few countries but also needs to be assisted by the LDCs and developing countries for a holistic development in the field of IP.
As well as to argument the steps of fulfilling SDGs.
The Delegation of India reiterates that these steps of providing child boxes and other features may be carefully analyzed and respected under the eye of the relevant national laws as applicable.
The Delegation would like to take this opportunity to inform.
All the Member States that India, Indian IPO is currently also exploring the options of implementing AI and block chain for the patent-related procedures for faster, secure and smooth processing of IP applications.
I do not see anyone else wishing to speak on this topic.
Cameroon, you have the floor.
Thank you for giving me the floor and I would like to thank Japan, the UK, France, Spain and WIPO for their presentations and the Secretariat as well for having organized this exchange session now during.
All this work we ask that there are interesting tools which have been developed by these countries.
And that resorting to artificial access to mutator in our systems taking into account the tremendous divided technology exists the SCP should continue.
Oh TO have these expenses of experience take place and step at the end come out with an out and document which is it would be a compilation of all the mechanisms take into account what was said my concern.
Is this subject in four points, one of the costs, number one, that this change imposes or entails, can we have the technologies use, this use would enable us to reduce the
Dramatically the timeline so for a.
Be able to provide patents which are the marches from mistake that we have and then taking into account the exaggeration to change in the system with artificial intentions.
Our deficient is this an advantage or not, it is a question, thank you.
The Africa Group would like to thank all Delegations that have shared their national experiences on the use of IP in patent examinations.
We also thank the Secretariat for its presentation on the use of a various processes in the organization and also extending these tools to other UN agencies, the presentations have been very informative.
From the presentations by Delegations we take note that almost all of those Patent Offices that you say I.
It contributes to the efficiency of the patent grant processes and ultimately the quality of a patent granted, however, we take note that AI is deployed in these countries for limited tasks.
Including prior art such and language translation the human element is still also very of great importance in these countries.
Was it doing that very few countries have so far deployed an eye for their patent girl processes.
And the limited application of I at this stage of the late format, we urge for cautioning discussions on AIR in the SCP.
Which we believe that the SCP should develop a balanced work programme that will ensure that the already existing dictor divide is not.
The group would therefore suggest that the Secretariat should undertake a cost benefit analysis of the use of the IP for Patent Offices.
In particular, we suggest that the survey should be undertaken by the Secretariat to determine the irreadiness or lack of readiness for use of IP patent offices in Developing Countries and Least Developed Countries.
The study should be able to identify existing gaps to enable WIPO to determine what kind of work.
Should be undertaken to assist Member States to beat their capacities to ensure that no Member States left behind thank you for your kind attention.
Thank you.
Okay. Thank you very much, if there are no further comments, we will move on to the next item.
The section of this agenda item which is the report on the technical assistance activities by the Secretariat relating to opposition systems and other administrative revocation mechanisms so I hand over to this.
Secretariat. Thank you,
Thank you,
Thank you, Madam Chair, the document which is relevant to
This subject matter is SCP/31/4 entitled report on WIPO's technical assistance activities relating to opposition systems and other administrative revocation mechanisms.
You may recall that at the previous session we had an agreement that the first.
The WIPO Secretariat should prepare this document introducing our activities on the technical assistance ready to this matter.
And although the document is relatively short, we try to capture the different aspects of the technical assistance activities that we have been conducting so very
In a very concise manner the areas that are touched upon are the.
1, the legislative and policy advice area, second, various activities held under the WIPO Academy.
And then third one of the projects which is conducted in the Asian region.
And then the third-party observation system under the Patent Cooperation Treaty and then finally the general provision of information.
On the topic of opposition system and other location and administrative location mechanisms.
In our publications or on the on the WIPO website, thank you very much.
Is anyone have any comments?
On this document.
Thank you, everyone, so if there is no comments on that document, we will close the
The lunch now will pick up on discussions on proposals by Member States and quality patents and opposition systems up to lunch and then continue with our agenda item on patents and health.
Thank you. Any announcements? Sorry. Croatia. Thank you, Madam Chair. CEBS will meet at 2: 30 in the Bilger Room. Thank you.
Any more analysis, no great have a lovely lunch, everyone thank you.
[[Powered by WIPO Speech-to-Text© Artificial Intelligence]]