The World Intellectual Property Organization (WIPO) published recently its first report in a new flagship series: WIPO Technology Trends, this first edition features Artificial Intelligence (AI) as the theme.
If you need help with Artificial Intelligence or intellectual property protection, we can help you.
Insights and trends in Artificial Intelligence
The main insights and trends in Artificial Intelligence techniques and application fields, and patenting activity are summarized below.
AI-related inventions are booming, shifting from theory to commercial application:
Since AI emerged in the 1950s, innovators and researchers have filed nearly 340.000 patent applications and published over 1.6 million scientific publications.
Notably, AI-related patenting is growing rapidly: over half of the identified inventions have been published since 2013.
Some areas of Artificial Intelligence are growing more quickly than others:
Machine learning techniques
Machine learning is the dominant AI technique disclosed in patents and is included in more than one-third of all identified inventions (134.777 patent documents).
The machine learning techniques revolutionizing AI are deep learning and neural networks, and these are the fastest growing AI techniques in terms of patent filings.
Among AI functional applications, computer vision, which includes image recognition, is the most popular, followed by AI for robotics and control methods.
Many Artificial Intelligence patents include inventions that can be applied in different industries:
AI-related patents not only disclose AI techniques and applications, they often also refer to an application field or industry. Twenty application fields were identified in the report, these include in order of magnitude: Telecommunications, Transportation, Life and Medical sciences, Human-Computer Interaction, Banking, Security, Agriculture and Networks (including social networks, smart cities and the Internet of Things).
Companies dominate patent activity
Companies, in particular those from Japan, USA and China, dominated patenting activity:
Companies represent 26 out of the top 30 AI patent applicants, while only four are universities or public research organizations.
IBM has the largest portfolio of AI patent applications with 8.290 inventions, followed by Microsoft with 5.930, Toshiba (5.223), Samsung (5.102), NEC (4.406), Fujitsu (4.245), Hitachi (4.180), Panasonic (4.175), Canon (3.940), Alphabet (3.800), Siemens (3.550), Sony (3.500), Toyota (2.850) and NTT (2.730).
Geographical distribution of patent filings
Most filings are made at the U.S. Patent and Trademark Office (USPTO) (150.000 patent applications) and Chinese patent office (137.000), followed by Japan (85.000), the European Patent Office (EPO) (54.000), Korea (37.000), Germany (26.000), Canada (13.000) and Australia (12.500).
Cooperation in AI research is limited, but so is conflict:
In many cases, organizations that cooperated in research are credited as co-assignees on patent applications. However, none of the top 20 applicants share ownership of more than 1% of its AI portfolio with other applicants.
Overall, the amount of litigation identified in the report is relatively low (less than 1% of patents being litigated), which may be due to the fact that products have now yet come to the market and infringement may be difficult to prove.
Source: www.wipo.intRead More
Artificial Intelligence and GDPR
The interaction of Personal Data Protection and Artificial Intelligence (AI) becomes particularly interesting when issues arise from the use of personal data with AI.
General Data Protection Regulation (GDPR)
The new General Data Protection Regulation (GDPR) of the European Union (EU), which entered into force on 25 May 2018, aims to give control to citizens of and residents in the EU over their personal data.
Regarding Artificial Intelligence, in particular, GDPR aims to create transparency rights and safeguards against automated decision-making, meaning decisions that are made by machines when personal data is used.
In essence, GDPR states that:
- When companies collect personal data, they have to say what it will be used for, and not use it for anything else.
- Companies are supposed to minimize the amount of personal data they collect and keep, limiting it to what is strictly necessary for those purposes stated. They also are supposed to put limits on how long they hold that data, too.
In short, companies must tell people what data they hold on them, and what’s being done with it.
- Companies should be able to alter or get rid of people’s personal data if requested.
- If personal data is used to make automated decisions about people in an AI system, companies must be able to explain the logic underpinning the algorithm used for the decision-making process, i.e., the general functionality of the automated system.
In particular, Article 22 of the GDPR grants individuals the right to contest a completely automated decision if it has legal or other significant effects on them.Read More
The three levels of Artificial Intelligence (AI) can be defined as follows:
- Artificial Narrow Intelligence (ANI) refers to a computer´s ability to perform a specific (single) task.
- Artificial General Intelligence (AGI) which is capable of transferring knowledge from one domain to a new domain, e.g. when a computer program can perform the same intellectual task as a human being.
- Artificial Super Intelligence (ASI) which is theoretically capable of surpassing human intellect.
Today, most experts would agree that we are seeing tangible results from ANI only. AGI is at least two decades away from being perfected (only scientific studies but no more), and ASI is even farther off (still kind of science fiction).
There is a diffused sentiment that the trend is that software will widely become AI, which in turn will become “Super Software” implemented everywhere, in every field of technology, well beyond ICT.
Key elements to be considered when drafting/prosecuting a patent application in the field of AI are the following:
- The technical effect(s) of the invention should be explicitly defined by means of technical features.
- The AI invention must be a technically implemented.
- The AI invention must be applied to a field of technology.
Theoretical and futuristic aspects include the following issues:
- Who is the inventor, a person or AI?
- Reverse engineering will be more and more difficult: How to detect infringement?
- Who is the skilled person for assessing inventive step, a person, AI or a combination of both?
The potential of Artificial intelligence is enormous. The interconnection of A.I. and art is a relatively less explored field, but in this regard the consequences for intellectual protection law are unpredictable and equally interesting.
In 2016, a network of Dutch museums presented a portrait created by a computer that analysed hundreds of Rembrandt paintings, to finally produce a new artwork in the style of the Dutch artist and that looks exactly as if it had been made by Rembrandt himself.
In the same year, a computer software wrote a brief novel that was admitted into a literary prize in Japan, although it didn’t make it to the final round.
Deep Mind, a Google company, has developed a program that creates original new music by listening to old recordings. The first computer-generated musical debuted in London as early as 2015.
While bots have been present in the creation of work of arts since the 1970s, the difference with modern A.I. lies in the fact that, in the past, the human programmer still had a relevant creative input and the machine merely executed, or reproduced his ideas adding an element of randomness. Today, evolved A.I. robots – due to the developments of machine learning -are capable of behaving in a sophisticated way which is almost as indistinguishable from human intelligence and are capable of unpredictable, autonomous decisions.
As copyright law protects original works of art, and the works produced by robots are certainly original, who owns the copyright to those artworks?
The consequences for copyright law
The question is relevant as there may be substantial commercial rights attached to the copyright of a work of art. Who is to benefit from the commercial rights? The developer of the software? What if the activity cannot be tracked down to a developer in particular, but to a whole company – like it usually happens in case of very sophisticated programs? Or is the machine itself to be considered as an author – but again, in that case, who holds the economical rights? And what if, on the contrary, the final work cannot be considered as attributable to anyone – and therefore can be reproduced, modified and commercialized freely?
Most jurisdictions, included South American jurisdictions, are unequipped to deal with artificial intelligence from this perspective. In Mexico, for instance, the concept of copyright is always linked to a person.
Art. 12 of the 1997 Federal Law on Copyright, states that “The author is the natural person who has created a literary or artistic work.”
In other words, the existence of an individual identified as the author is essential; and by the same reasoning, for instance, corporations in Mexico cannot be authors, although they may be the copyright holders.
The same is true for Brazil, where section 11 of the law 9.610/ 1998, establishes that “The author of a literary, artistic or scientific work is the natural person who created it” – negating the status of creator both to companies and to the software itself.
On a global perspective, EU courts have been behaving similarly, in the sense that they attribute the rights only to a human author – see for instance the Court of Justice of the European Union (CJEU) in the decision C-5/08 Infopaq International A/S v Danske Dagbaldes Forening, where it stated that originality must reflect the “author’s own intellectual creation”, implying that a human author is necessary for a copyrighted work to exist.
The United Kingdom adopts a more blurred stance on the matter, as UK copyright law, section 9(3) of the Copyright, Designs and Patents Act (CDPA), states:
“In the case of a literary, dramatic, musical or artistic work which is computer-generated, the author shall be taken to be the person by whom the arrangements necessary for the creation of the work are undertaken” and this needs to be paired with section 178 of the CDPA defines a computer-generated work as one that “is generated by computer in circumstances such that there is no human author of the work”, admitting, in theory, that computer-generated works may exist. It just needs to be defined who is responsible for the “necessary arrangements”, whether the programmer or the machine itself, which is still under debate in the UK.
Quickly developing technologies have put a strain on most legal systems, and South America needs to keep up with its advancements.
The European Patent Office (EPO) held on 30 May 2018 a conference on Patenting Artificial Intelligence (AI).
Koen Lievens, Director Operations at the EPO, presented as keynote speaker the topic “How does the EPO deal with the challenges of AI in patents”.
The key concepts are the following:
Computer-Implemented Inventions would also apply to the inventions relating to AI, therefore to examine AI inventions the EPO two-hurdle approach should be used.
Are AI and Machine Learning (ML) just mathematical methods?
According to Articles 52 (2) and (3) EPC, mathematical methods as such are considered non-inventions and therefore not patentable.
Algorithms used for the purpose of, among other things, classification, clustering, regression and dimensionality reduction would be mathematical method as such and therefore not patentable subject-matter.
Algorithms applied, among other things, to data of technical nature, parameters of technical nature and trainable based on training data would not be considered mathematical method as such and therefore would not be excluded from patentability according to Articles 52 (2) and (3) EPC.
To overcome the second hurdle the mathematical method (steps) should contribute to the technical character of the invention.
There are two dimensions to contribute to the technical character of the invention:
1)The AI and ML method (steps) is adapted to a specific technical implementation, 2)The AI and ML method (steps) is applied to a field of technology.
In the case 1), the AI algorithm should be specifically adapted to a specific technical implementation, and, furthermore, the AI design should be motivated by technical considerations of the internal functioning of the computer.
Usually, generic technical implementation, mere programming or algorithm being merely more efficient than in prior art will not be sufficient to contribute technical character.
In the case 2), the AI and ML method (steps) should serve a technical purpose by means of a technical application, i.e., to solve a technical problem in a technical field. AI technical application fields would be image processing, speech processing, fault detection/predictive maintenance, medical analysis, self-driving cars, etc.
A generic statement like “controlling a technical system” would not be sufficient to contribute technical character.
Sources: www.epo.orgRead More