Increasing number of AI patents

Today, artificial intelligence, for which the acronym AI is commonly used, has numerous practical applications in several technical fields. Intelligent software can be used, for example, to automate routine work, make medical diagnoses in medicine, understand or interpret speech or images, control robots, or allow vehicles to move autonomously.

Using so-called machine learning models, computers are able to learn from experiences and understand the world in terms of a hierarchy of concepts, with each concept defined by its relationship to simpler concepts. Because computers themselves acquire the knowledge from experience - the ‘training’ of a model - it prevents human operators from having to formally specify all the knowledge the computer needs to perform certain intelligent tasks. The hierarchy of concepts allows the computer to learn complicated concepts by building them from simpler concepts. An important subfield of artificial intelligence is the so-called artificial neural networks. These are brain-inspired models used to make accurate data-driven decisions. Such models are often very carefully tailored to a specific practical situation.

AI inventions can be protected by patent if they solve a technical problem in a technological field. For patent protection, it is important that the invention has a technical character. After all, the features of the invention that make no technical contribution cannot contribute to inventive creativity, which is a prerequisite for patent protection.

An example of an invention of a technical nature is an AI training method that causes a neural network to converge faster or use a smaller data set. This has the technical effect of allowing the computer to work more efficiently, and thus the improvement is eligible for patent protection as an invention. Another example is the use of a neural network in a heart monitoring device that provides a technical contribution for the detection of irregular heartbeats. Such an AI invention specifically designed to recognize patterns of a signal over time and to recognize deviations from those patterns has an adequately defined technical purpose, and therefore may be patentable.

Known machine learning models as such, such as neural networks, are considered standard technology. For example, if a neural network is merely presented as a component in an existing application, the invention is usually not considered inventive. The situation is different if serious consideration is given to how the application and/or intelligent component should be modified to achieve a working version. Solving technical problems in a specific technical implementation can be an indication of inventive creativity.

V.O. sees that patent applications with AI inventions in various technical fields are becoming more common. We expect that this trend will certainly continue in the future. Moreover, we see that this trend is even occurring in technical fields that were not traditionally considered computer-oriented. For example, the European Patent Office has found that about 40 percent of new patent applications filed in the healthcare sector include an AI or machine learning aspect.

We are happy to help you protect your future AI innovations.

Our experts will be happy to inform you about all the ins and outs.