Prometheus made its presence felt at the conference on “Artificial Intelligence- Intellectual property policy considerations” organised by the United States Patent and Trademark Office (USPTO) at its headquarters in Alexandria, Virginia. The conference was attended by our US Operations Head Ms Soujanya Sikha. She extensively interacted with the speakers, exhibitors and fellow attendees on various issues related to Artificial Intelligence (AI), its application and issues related to patent eligibility of inventions involving AI.
Opening remarks by Director USPTO Andrei Iancu:
The importance of Artificial Intelligence (AI) for national, international, industry, academic and even kitchen table discussions was mentioned. He said “speech recognition, image recognition, search and optimization, and even early neural networks—were conceptualized as early as the 1950s; the significant breakthroughs are more contemporary.”
He spoke about the transition of AI applications from Sc-Fi to reality. He emphasized on adapting a long term strategy with reference to AI application. The contribution of academia and industry to provide techniques for innovating faster and in key areas was emphasized.
AI is used for presenting results in the form of pre search for obtaining relevant search results. He also mentioned that the policy makers should consider these issues (related to AI), as to how to assess patentability when little transparency with respect to algorithm is there.
Remarks by Dr Lynn Parker, Assistant Director for AI, White House official of Science and Technology Policy
There is an official US policy and a national security commission on AI. These systems are considered different from computer science based algorithms as the AI systems have the ability to learn. She focussed on the capacity of AI to increase the economic prosperity. A sudden interest in AI and its applications was attributed to faster processing of data. Though digitization is not new but dealing with big data in with accuracy and speed is where the AI scores brownie points.
The innovation ecosystem for AI in USA is comprised of Federal Government, Private sector and Academia. She pointed out certain regulatory challenges faced by AI when put to use like: driverless cars, retino-neuropathy, and drones etc. which erupt while granting patent protection.
EXCERPTS FROM VARIOUS SESSIONS:
IP law issues are creeping up and affecting the AI and its practical aspects. An overall argument especially from the engineers is that since the invention involves application of AI hence it’s patentable. The corporates believe in the emphasis to deeply dig into details to obtain the highest level of protection.
With reference to patentability another argument is that AI is all about software. Patentability of software issues automatically apply. The federal circuit decisions normally hold patent eligibility responsible for all rejections. The examination of AI applications has not changed and there are many 101 rejections. Buzz words related to AI are used a lot in patent applications but these are not practically implementable. With reference to 101 rejections it may be cited by the applicant that AI is not something which can be done by a person’s mind “per se” and there is a genuine application of it in the invention.
New examiner guidelines for software applications issued by the USPTO address the problem of some issues, but a detailed interpretation is required. Some panellists opined that subject matter eligibility is not a criterion at all. Guidance issued doesn’t have the clarity on what happens when we go to enforce a patent. Sometimes though a patent for AI granted, but they don’t justify execution for example facial recognition etc. Some arguments are that courts should not decide on the subject matter eligibility.
America is still working on grace periods arising out of pre AIA and post AIA. AI as an abstract idea and what is an inventive concept etc. are examples which have not been catered to while framing the guidelines for evaluating patent applications involving AI. About 80% of AI patent were rejected in 2015 and in 2018 the rejections dropped to 57%. The courts in the US have not defined the concept of an abstract idea with great clarity.
It is important to provide enough improvements to be cited in one’s specification through AI and tying that to claims. Check for any non-generic components put to use in the invention. Though non generic components may be argued from legal aspect but explanation must be given for their usage.
Corporate giants are opposing attracting excessive AI scrutiny in the name of algorithms, though they agree that IP protections are essential for providing incentive to the inventors. It is to be analysed whether current set of IP laws are accurate for the new technology in question? A Sui generisis system for Business methods and likewise for AI may be proposed. A post AIA concern is that AI benefits the big and inventors and corporate giants and the small inventors are at stake. China’s compromise on data privacy regulations was questioned.
With reference to copyrights, can writings of a computer be considered on par with the writings of a human is a big question. Granting copyrights is a total misfit for AI in the opinion of some attorneys. Microsoft spoke of EU providing a sui generisis protection for databases. Computerised reading may not always be considered as a fair use. Computer reading approach should be paid for the original creator. The argument to this is that computers can never be expressive as the author. A Sustainable economic pattern should be evolved for the authors.
In the session on Global perspectives with reference to AI and IP policies around the world the points discussed included; the EU issued guidelines in October, 2018 with reference to AI. There is no concept of fair use in Europe, but there is a list of exceptions. China has been a pioneer and has the largest number of patents to its credit in AI. China has been particular in enhancing AI to cover technological gap and to enhance the economy. SIPO IN 2017 issued guidelines to speed up applications in priority industries recognized by state to be done for examination including AI.
Japanese Patent office issued in 2018 a set of contract guidelines issued while making contract related to AI.
Singapore was described as a case study where the patent laws are a combination of policy, politics and X-factor. Pragmatism was considered as a part of the existential policy of the country. Singapore is the 2nd largest investor in USA and largest in China. AI immediately implementable plan was released in Davos, 2019 during world economic forum by Singapore (First of its kind in Asia). The speaker on Singapore’s policy clarified that it does not have 101 issues and it believes in the concept of “let a thousand flowers bloom”. It was also opined that AI should not be considered as a substitute to machines. A further discussion involved TM issues, image variations in logos, counterfeiting etc. which are being extensively targeted by the implementation of AI.