Conference report: “Can robots invent and create? A dialogue between Artificial Intelligence and Intellectual Property”

Can robots invent and create? This was the key question at the heart of a 2-day conference that was held in Alicante on 14 and 15 March last.

Katfriend Vicente Zafrilla (Doctoral Researcher at the Max Planck Institute for Innovation and Competition in the framework of EIPIN Innovation Society (H2020- Marie Skłodowska-Curie Action)) has prepared a report, also thanks to the help of Myriam Christmann, Raúl Ruiz, Rubén Cano and Carlos Muñoz Ferrandis.

Here's what Vicente writes:

L to R: Alejandro Roca, Giancarlo Frosio,
Manuel Desantes and Ilja Rudyk

As IPKat readers might be aware, in March Alicante became the capital of the Artificial Intelligence and Intellectual Property for a couple of days, thanks to the congress “Can robots invent and create? A dialogue between Artificial Intelligence and Intellectual Property” organized by Fide and hosted by Universidad of Alicante, under the coordination of Javier Fernández- Lasquetty (Director), Prof. Aurelio López-Tarruella (Director) and Prof. Manuel Desantes Real (Academic Coordinator).

Day 1

The Congress kicked off with the keynote address of Dr. Nuria Oliver who pointed out that, although AI could overcome human limitations, such as those caused by stress, fatigue or conflicts of interest, it is also possible that it perpetuates certain biases.

To prevent AI from resulting in a tyranny of experts - and even in a tyranny of algorithms - AI must be developed based on a humanity-centric approach, and legal professionals must contribute to shaping this environment in accordance with these ethical, and desirable, principles.

The second session “Artificial intelligence and Intellectual Property, elements for a dialogue” provided a comprehensive overview of some of the aspects that were discussed during the subsequent sessions. Ilja Rudyk presented the increasing number of patent applications in the AI field at the EPO and anticipated some of the EPO approaches to AI-related patents.

Dr. Giancarlo Frosio offered a full picture of the challenges that AI poses for the IP environment. According to him, such challenges are related to inputs for AI (data), outputs from AI (authorship, inventorship or cultural diversity, among others), Digital Rights Management, infringements and liability.

The session concluded with the presentation of WIPO´s recent report “Technology trends: Artificial Intelligence” by Alejandro Roca who highlighted the importance of empirical evidence for decision making.

The floor was opened for session 2 “Artificial Intelligence & Authorship. Challenges for the copyright system”, during which Prof. Raquel Xalabarder discussed wherther granting no copyright at all to AI works could lead to unbalances in the system and pervert some of the incentives which are at the basis of the copyright system. In addition, she pointed out that historically technologies have faced similar situations with regards copyright protection, as it happened in the cases of photography or phonograms.

IPKat's own Dr. Eleonora Rosati focused on the interaction between copyright and AI in the field of Text and Data Mining (TDM). Certain TDM techniques require performing a series of actions – namely access to the content, extraction and copy – which potentially conflict not only with copyright, but also with related rights, data protection or sui generis rights. Against this background, the TDM exception in the DSM directive was initially limited since its scope would only encompass research institutions. Nevertheless, the new wording of Art 3.a of the Directive (now Article 4) opens the door to TDM performed by businesses insofar as the right holders have not expressly reserved the right.

Prof. Bernt Hugenholtz considered that the key aspect is not to get to the right answers, but to ask the right questions, stating that asking whether AI can be an author is the same thing as debating whether kangaroos can compete in the Olympics. AI can indeed provoke a shift in the focus of copyright protection: from the act of creation to either its conceptualization or to the end of the creative process as such. Nevertheless, protection – if any - for AI-generated works needs to be balanced against the overarching goals that such protection aims to achieve.

Session 3, titled “Challenges for the patent system. Inventions developed by Artificial Intelligence: breaking the boundaries”, started with the presentation of Dr. Alberto Casado, who pointed at clarity for the protection and litigation of AI as a key factor to adapt the XIX century-created patent system to a reality in which software is more and more present in inventions.

Doris Thums analysed four AI-related critical areas for EPO practice: first, as per inventorship, the state of the art still requires human intervention; second, AI-based inventions could be protected as computer implemented inventions; third, although more clarity is desirable for the disclosure requirement – in relationship with “black box” AI processes – all applications to date fulfilled the requirement; and fourth, AI may potentially affect the “skilled person” standard. To conclude, the speaker affirmed that the EPC and the EPO are ready for the future.

Aurelio López-Tarruella and
Anke Moerland

Finally, Prof. Noam Shemtov arrived to a similar conclusion: the patent system is suitable to face current challenges posed by AI, after explaining that human intellectual domination of the process suffices for the purposes of attributing inventorship and that the current patent incentive theories can be challenged if inventorship is attributed to AI.

The following session, “Artificial Intelligence and technical innovations. Challenges for the patent system. Computer implemented inventions. Trade secrets and Artificial intelligence”, started with an overview from Julio Raffo concerning patenting trends in the robotic field, based on WIPO study “Breakthrough technologies – Robotics, Innovation and Intellectual Property”.

Then, Jean-Marc Deltorn focused on the practicalities of AI-related patent examination at the EPO, by addressing two “hurdles”. While the first, related to the technical character of the invention, can easily be overcome thanks to the “any-hardware” approach, the second, linked to the inventive step requirement, requires the applicant to show that the algorithmic feature performs a technical function within a specific implementation in a field of technology.

The last speaker of the session, Emmanuel Gougé, switched the topic to trade secrets, stating that trade secrets are a relevant legal tool to safeguard critical innovations in the AI field, highlighting its complementarity with other IP rights – patents and copyright – as a key factor for its use in IP protection strategies.

The last session of Day 1, “Artificial intelligence and emotional innovations. Will Artificial Intelligence change the trade mark system?”, tackled the not-that-obvious relationship between trade marks and AI.

Lee Curtis discussed how AI will transform the buying model from a “shop it, then ship it” model to a “ship it, then shop it” in which AI will interact with trademarks in a different fashion than humans do: aspects such as imperfect recollection or likelihood of confusion are not likely to be that relevant in cases where AI takes the purchasing decision. In addition, issues related to AI liability are likely to arise sooner or later.

Dr. Anke Moerland explained the attendees how well-suited the trademark system is to tackle trademark infringements within virtual and augmented reality environments in cases such as likelihood of confusion, dilution, tarnishment or comparative advertising.

The session ended with a speech by Erling Vestergaard, who showed how trade mark infringers use AI-based systems to detect and buy domain names abandoned by trade mark holders to automatically set up, in less than 24 hours, platforms to sell counterfeits and how these systems are responsive to enforcement actions. His speech ended by highlighting some of the initiatives that the EUIPO is promoting to use AI and blockchain as a mean to detect trade mark infringers and act against them.

Day 2

Day 2 started with a lively debate between Prof. Ricard Martinez and Prof. Andrés Pedreño on the interplay between new technologies and human rights. The former advocated for a regulation that would guarantee the protection of users’ rights and stressed the role of privacy by design principles to help business and legal professionals to respect fundamental rights without hindering innovation.

Prof. Pedreño, on the other hand, noted how overregulation negatively affects innovation, particularly in the case of SMEs which are less able to bear compliance-associated costs. He proposed a more coherent, innovation-aligned and harmonized legal framework.

In session 7 Prof. Josef Drexl presented the links between “Artificial Intelligence and non-personal data: the case of Industry 4.0” and discussed the suitability of the current IP regime to protect non-personal data. Prof. Drexl tackled whether the focus of protection must be placed on “creating information” rather than on “exploiting raw data”. Existing IP tools such as copyright – that can protect the program but not the data itself – or sui generis database rights – which may create barriers for data sharing - are poorly suited to adapt to this environment. However, it does not mean that AI will lead to the end of IP, neither ground an eventual reform of it, reform that in any event should be done at the light of existing protection tools and after evaluating its impact on competition as the most important driver for Industry 4.0.

The following session, “Techno Round Table”, gave attendees a taste of business and technological approaches to AI. Miguel Arjona presented some of the potentialities of technologies such as Object Recognition, Deep Learning Algorithms or Generative Adversarial Networks.

Then Pablo Haya focused on Natural Language Generation and its uses to create content such as tales, articles or a TV presenter speech.

The third speaker, Angel Sevillano, presented four case studies of the use of IBM´s Watson: to analyse perfumes' ingredients; to create the song “Not easy”; to design a fashion collection in collaboration with Custo Barcelona; and to compete in a debate contest.

Finally, Luis Ignacio Vicente del Olmo, discussed ssome of the ethical, strategic and technological challenges that Europe is facing nowadays. These include the need for a European response to US and China leadership in AI, non-suitability of traditional ICT licensing models for the current cross-sectorial dynamics and lack of harmonization.

The congress ended with a keynote address by Tobias McKenney who noted how fragmentation of copyright and lack of harmonization pose difficulties to the control of copyright infringements on YouTube and discussed some of the results of the paper “How copyright law creates biased Artificial Intelligence”.