1. 30 highly skilled professional manpower in semiconductors start public office.
- Securing a super gap in semiconductors is supported by using highly experienced people with extensive field experience.
- (Successful candidate A) was very attracted to a job opening with the intent to hire a person having extensive field experience in semiconductors as a patent examiner and thought the job as a valuable opportunity to make use of the know-how mastered as a technician and diverse experiences.
- (Successful candidate B) experienced the issue of technology leak and the importance of a patent while watching the reality that many of Korean semiconductor industrial colleagues transferred abroad and thus applied to play a part in maintaining the position of the super-differential power.
- (Successful candidate C) had great regret as a semiconductor-related person when hearing that it took about 2 years for a semiconductor technology patent from being filed in an application to being registered as a patent and thus hoped to be hired as a patent examiner to keep a national semiconductor industry in the global semiconductor hegemonic competition.
On February 23, 2023, KIPO published a list of 30 final successful candidates to be hired as patent examiners of the professional term system in the semiconductor area.
This employment was carried forward as a part of the national tasks to prevent core technology from being leaked abroad by the overseas job transfer of the skilled manpower in the Korean semiconductor area in the competition for technological hegemony and to support and secure the semiconductor super gap by using the manpower’s extensive field experience and knowledge in patent examination.
There initially were internal and external concerns that the support for a semiconductor expert would be difficult due to a low salary compared to a private sector and the lack ofguaranteed retirement age. However, as a result of receiving applications, 175 people applied showing a high competition rate of 6:1 which was unusually high in light of the recruitment of an examiner for a professional term system showing a competition rate of usually about 2~3:1.
To grasp the depth of a candidate’s competence, KIPO appointed an external expert by field of specialization as an evaluator and conducted a 2 month intense document examination plus individual interviews, to finally selected a total of 30 successful candidates.
From the successful candidate status, the oldest one is 60 years old, the youngest is 41 years old and the average age is 53.8 years old.
Semiconductor highly-skilled professionals as selected are familiar with the latest technology trends with ▲ a semiconductor-related average career of 23 years and 9 months, ▲a master’s and doctorate degree retention rate of 83%, and ▲ an incumbent rate of 90%.
After appointment, the successful candidates will be arranged at departments by detailed technical field, such as semiconductor layout, process, materials, etc., through a new examiner education, etc., to execute patent examination work. They will also take a close-up guidance (mentor-mentee) for about 2 years to cultivate examination ability.
A number of Korean companies have been highly interested in this recruitment by KIPO. A person of the personnel department of an “A” company visited KIPO and hoped that this recruitment would be promoted continuously not on a one-off basis since it would be a government policy to be substantially helpful in the semiconductor companies. A person of a “B” company suggested that this kind of recruitment would be needed to expand to other cutting-edge technology areas as well as the semiconductor area.
According to KIPO, considering that 150 people (86% of the candidates) were from semiconductor companies and 4 people were experienced in overseas companies and supported for domestic U-turn, this recruitment has been confirmed as having measures to prevent technology leakage.
2. The technology competition of “hyper-scale AI”, the core of Chat GPT, has been intensified.
- The number of patent applications related to hyper-scale AI grown by 28 times for the last 10 years.
- Korean companies including Samsung in 1st place and LG in 10th, etc. have led the patent applications
While ChatGPT, a conversation AI which was published on November 30, 2022 by the company, OpenAI, has been a social issue, competition for a patent(s) to occupy in advance the hyper-scale AI technology based for ChatGPT has been intensive.
According to KIPO, the number of patent applications related to hyper-scale AI, which were filed in IP5 (the Republic of Korea, US, Japan, China and Europe), increased by about 28 times over the last 10 years (2011~2020) (from 530 in 2011 to 14,848 in 2020, with an average of 44.8%).
Notably, the number of these applications over the latest 5 years (2016~2020) increased by an average of 61.3%, showing the increase in the number of the applications was growing faster. This is considered as resulting from the active research of AI since the impact of AlphaGo in 2016.
From the view of the applications by applicant’s country, 15,035 applications (35.6%) were filed by US, 13,103 (31.0%) by China, 4,906 (11.6%) by Japan, and 4,785 (11.3%) by Korea, taking 4th place with a slight difference.
However, in terms of an annual average increase rate, Korea (annual average of 89.7%) and China (annual average of 79.3%) showed a steeply rising trend.
Specifically, Korea filed only 6 applications in 2011 but 1,912 in 2020, showing a rapid growth of 319 times. Since 2019, the annual number of the applications filed by Korea has passed that by Japan.
Based on the hyper-scale AI technology development trends, ① data generation technology was 69.3%, followed by ② learning model technology (25.8%) and ③ specialized service technology (16.4%).
The number of the patent applications related to ‘learning model’, which is the core technology of hyper-scale AI rapidly increased (annual average of 75.9%) and notably it increased by 126.3% every year over the last 5 years (2016~2020) indicating that the research and development in the learning model technology area were active.
The major applicants were Samsung in 1st place (1,213 applications, 2.9%), IBM in 2nd (928, 2.2%), Google in 3rd (824, 2.0%), Microsoft in 4th (731, 1.7%) and Baidu (572, 1.4%), showing that the global big-tech companies took the top spots.
As Korean companies and research institutes, Samsung was in 1st place, LG in 10th (384, 0.9%), Stradvision in 25th (209, 0.5%), ETRI in 36th (157, 0.4%), KAIST in 66th (80, 0.2%). Not only big companies but also applicants in various fields, such as small and medium companies and venture companies, research institutes, universities, etc., had an internationally competitive power.
The patent applications of hyper-scale AI were regarded as being centered on ‘companies’ (78.7%).
Specifically, US (91.2%) and Japan (95.4%) had high proportions of companies. Korea had an increase in the number of the relevant application filed by companies, from 50% in 2011 to 73.6% in 2020.
3. 450,000 cases relating to medicine experimental data and AI learning data are available free.
- From February 13, 2023 these are available in a patent information utilization service of KIPO.
KIPO open free a total of 450,000 cases relating to medicine experimental data and AI learning data to extract the experimental data, which are included in the patent publications, through a patent information utilization service (KIPRISPlus, plus.kipris.or.kr) from February 13.
The patent information utilization service (KIPTRISPlus), which is a data open distribution network (platform) operated by KIPO, provided the major intellectual property (IP) (patent, trademark, design) publications of 13 countries at home and abroad and the data products such as IP administration information.
Recently, the AI learning data for multilanguage translation, image search, etc. have been open such that a total of 115 kinds of data products are open in a file or OpenAPI (open application programming interface) format.
The medicine experimental data which have been open are the information in which a medicine ingredient name, an experiment method, an experiment value, etc. are built as baseline data (database) by processing and analyzing a table in an image format, which is included in a medicine patent publication.
The AI learning data include ① image classification information to extract only a table format from experimental data in various image formats, such as table, graph, chemical formula, etc., ② table structure (columns x rows) information to accurately grasp table data, and ③ experimental data attribute classification information to automatically classify an ingredient name, an experiment value, etc. used for an experiment.
When using the medicine experimental data and AI learning data, the ‘table experiment data’ in the image format included in a patent, thesis, etc. are converted and extracted in analyzable data in the text format.
Based on these converted and extracted data, IP service companies are expected to develop the extracting and utilizing services of experiment data in patent publications, and the relevant companies and research institutes are expected to freely analyze and utilize experiment examples, comparative examples, etc. including patent publications so as to be used in research and development of vaccines, new medicines, etc.
4. A trademark partial rejection system and a trademark re-examination request system are in effect.
- A positive administration for a trademark application applicant will be helpful in securing a trademark right.
According to KIPO, a partial rejection system is applied to a trademark application filed from February 4, 2023 and an applicant is able to request an examiner for a re-examination regarding a rejection decision of a trademark registration.
This newly enforced partial rejection system is to reject only a designated good having a rejection reason, among the designated goods in a trademark application.
Previously, when a part of the designated goods that an applicant wants to be registered was rejected, the total designated goods could not be registered unless the applicant deleted or amended the good(s) having the rejection reason.
However, as the partial rejection system is in effect, when only a part of the designated goods described in a trademark application is rejected, the other goods having no rejection reason can be registered even if an applicant does not take any separate measures, such as deletion of the relevant rejected good(s).
Therefore, this partial rejection system is considered to be helpful for an individual or small/medium company applicant who is not familiar with the trademark application procedures and systems and is unable to properly respond to the notice by the examiner due to time and cost issues.
In addition, previously it was possible to file an appeal regarding the whole of the goods rejected. However, it is now possible to file an appeal regarding only part of goods rejected and it is possible to withdraw part of goods even though the appeal is requested, to improve an applicant’s convenience.
When a rejection decision of a trademark application can be simply solved by an amendment to goods, etc., the re-examination requesting system makes it possible for the application to be re-examined by an examiner, thereby expanding the chance for the applicant to overcome the rejection decision.
Previously, since it was possible to proceed with appeal procedures against an examiner’s rejection decision by only an appeal against the rejection decision, the appeal should be filed even in the case where a rejection reason could be easily overcome by amending a part of designated goods only.
5. An IP examination will be faster and more accurately with AI.
- KIPO published an IP administration renovation implementation plan (roadmap) using AI technology.
KIPO made the year of 2023 as the first year for IP administration digital renovation using new technology, such as AI, etc. and established an IP administration renovation implementation plan (roadmap) (2023~2027) using AI technology .
Although the number of IP applications including patent applications has increased, since it is difficult to increase examination manpower and the scale of prior art to be searched by an examiner has continuously increased, overall examination environments have been harsher. The number of IP applications filed over the last 10 years increased by 49.3%: whereas, the number of examiners increased by 19.7%.
To overcome this, KIPO confirmed the use of AI implementation plan (roadmap) to apply the AI technology to overall IP administration including examination and trial, etc., to be fully carried forward this year.
With a core goal to build ‘a world top AI-based examination and trial system’, this implementation plan (roadmap) provides 4 promotion strategies and 12 key promotion tasks.
First, a base using the AI technology is built in overall IP administration.
In this year, KIPO cooperating with a private company will develop an AI language model specialized in understanding and processing IP documents and build a Korean language of a foreign IP document.
These tasks to be ultimately based for examiners to efficiently and accurately search giant IP documents will be developed by reflecting the AI technology readiness level and service situations.
Second, a high-quality examination is supported by using AI-based technology.
By utilizing a Korean translation of a foreign IP document, an AI IP search range which is possible in only Korean IP documents will be expanded to US and European patent documents, etc.
In addition, at present an AI trademark search is possible for only a figure trademark made up of an image. However, research and development to expand an AI trademark search in a word mark starts in this year.
Third, the AI technology is introduced in a trial and a formality examination.
An automation system will be developed to be used in a formality examination to check procedural flaws in various documents to be submitted to KIPO.
To digitalize the trial system, the systems to submit and deliver trial-related documents will be totally reorganized this year and the introduction of an AI-based judgement decision search service will be carried forward next year.
Fourth, the application of AI technology will be expanded in customer consultation and IP data utilization.
An AI chatbot consultation function will be upgraded for a customer-focused complaint service. The quality and availability of IP data will be improved by building diverse experimental data in an image file, which are included in some foreign IP documents or domestic patent documents, as database (DB) in a text format.