Artificial intelligence

AI illustrative image showing code and connections

Artificial intelligence (AI) is reasoning and decision-taking by machines rather than humans or animals. For example, in the past technology might have been used to monitor a process, but a human operator would still take the decision as to what parameters to change, even if technology was then used to implement those changes. AI effectively takes the human out of the picture. With advanced sensor technologies providing more information than a human brain can handle, AI can receive and interpret vast data volumes and respond appropriately according to the algorithms and parameters set by programmers.

AI illustrative image showing electrical circuits in the outline of a human as well as code snippetsAI is leading to the automation of screening of clinical samples, for example, faster and more accurately than human operators. Traffic volumes can be monitored by AI, leading to traffic signals being adjusted to compensate - all without human intervention. AI also concerns machine-learning, where a computer constantly assesses its own performance against feedback so as to do a better job next time - machine translation is a classic example of this.

The last decade has seen processors become smaller and cheaper allowing for their integration into everyday objects. Whole systems can be integrated on a chip complete with networking capabilities. This technology has allowed the development of physical object networks ("Internet of Things" -  IoT). Enabling technologies such as IoT and AI amongst others have permitted the development of objects operating autonomously depending on networked data to augment data gathered from their surroundings, e.g. displaying data in form of augmented or virtual reality.

A robotic hand engages with a digitial interfaceThe most evident application of AI to the patent world is prior-art search, which identifies statistically significant similarities between patent documents. As more patent applications are filed on-line it will be easier for machines to read them and find similar documents amongst the millions already in the databases. By the time a human examiner looks at the application, AI might already have given them a substantial headstart!

Artificial intelligence at the EPO

Progress in data analytics is of fundamental importance for extracting value from data. Analyses have long been performed by people supported by computers. However, the development of powerful diagnostic systems, including the performance of human-like cognitive functions by artificial intelligence (AI), is set to change this pattern. These tools can process vast amounts of data, and detect and interpret patterns that were previously impossible to calculate, identify, or even imagine. By making the interpretation of such patterns meaningful for machines as well as for humans, they enable machine prediction, diagnosis, modelling and risk analysis.

Grant Philpott"AI is clearly an area of rapid growth and immense opportunity for innovation. But the patent system will have to work very hard to ensure that it remains precisely that - an opportunity."

Grant Philpott, Chief Operating Officer for ICT

In May 2018, the EPO hosted its first conference on patenting artificial intelligence and was attended by more than 350 representatives of industry, academia, user associations, patent law firms, the judiciary, national patent offices and government bodies. The conference covered the challenges and opportunities of "Patenting Artificial Intelligence" with an aim to raise awareness and launch an open exchange of views on AI. In a series of presentations and panel discussions, participants discussed various solutions for how the patent system can provide applicants with a solid framework for patenting AI inventions.

Challenges and opportunities of patenting AI

AI illustrative image AI is related to patents in two ways. The first is that technology advancements can be protected by patents and the second is that AI can be applied to the patent space to eliminate inefficiencies. As a result of technology advancements, AI-related patent applications have surged over the recent years bringing with them eligibility challenges and legal litigations. There, challenges can be overcome by patenting those technical improvements and learning from the best practices on how AI can be used in the administration of IP rights.

Margarethe Zmuda, Senior Patent Attorney at Ericsson GmbH "There are challenges in each and every phase of the life of an invention: the evaluation and patent drafting phase, the patent application prosecution phase, and the enforcement phase."

Margarethe Zmuda, Senior Patent Attorney at Ericsson GmbH

Other solutions to these challenges include: the need to speed up the patent granting process; provide earlier publication of patent applications; possibly adapt the model itself so as to protect the invention or make substantial legislative changes (including patentability criteria) to address the challenges of AI.  

AI illustrative image Grant Philpott, the EPO's Chief Operating Officer for ICT, highlighted recent reforms implemented by the EPO to increase quality and efficiency, which have prepared the Office for the future: "AI is clearly an area of rapid growth and immense opportunity for innovation. But the patent system will have to work very hard to ensure that it remains precisely that - an opportunity."

The fast evolution of artificial intelligence and 4IR-related patent applications presents specific challenges. The EPO aims to continually improve and develop its processes and services to best fit with changing needs and ultimately to protect IP for the benefit and economic growth of Europe and society at large.

Quick Navigation