Patent applications must describe human effort, not just artificial intelligence (AI) 

Artificial intelligence (AI) can be successfully used in collaboration with inventions and innovations. It has been like this for a long time. At the same time, it can be demanding to describe the use of AI in a patent application in such a way that a patent is obtained on the basis of the AI use. Experience shows that such descriptions have changed over time. Today, the recommended practice is to focus on the human effort behind AI rather than on the artificial intelligence as such. 

Patent applications are fundamentally demanding documents to compile. They require technical descriptions and insight into legal matters. It will always be appropriate to seek expert help before a patent application is submitted. 

What is a patent and what are the requirements for a patent application? 

Patents are central elements in the management of intangible assets (IP) and are essential for protecting creativity and inventions of a technical nature. In order to have a technical solution patented, it must be new, not only in Norway, but throughout the world. It must stand out, and thus have inventiveness. In addition, it must be possible to reproduce the invention based on the information contained in the patent application. 

What is artificial intelligence (AI)? 

Artificial intelligence (AI) is an overarching term and encompasses all intelligent systems. There is still no clear definition of what concrete artificial intelligence is. What these artificially intelligent systems have in common is that they need an input in the form of training data, in the same way that humans need information in the form of sensory impressions in order for us to learn. 

A distinction is often made between two main categories of artificial intelligence. 1) Those models that are rule-driven and pre-programmed to behave in a certain way according to a given set of rules and 2) those models that use machine learning. 

Unlike rule-based models, machine learning models have no pre-programming that states that a given outcome is correct, but instead the model learns such rules through training on large datasets. 

Artificial intelligence and patent applications – then and now 

A few years ago, it was common to see patent applications that carefully described the functions and purpose of the invention, but often described the AI effort itself quickly and superficially. Roughly speaking, it was often described that the invention should have a function and a purpose and that artificial intelligence was used to achieve this. 

Today, this will hardly be enough. 

As long as the crux of the invention is represented by artificial intelligence, the invention is not patentable unless another, man-made solution is described in addition. If an innovation is to have the status of invention, it must represent measures beyond the use of artificial intelligence (AI). Or the AI solution itself must be described in such detail that it appears as an innovative element in the context. 

It is therefore necessary for the applicant to look more closely at the underlying, human effort that forms the basis for the AI-generated element in the patent application. Don’t just describe that AI is in use. 

Human effort behind AI-generated innovations 

As mentioned above, artificial intelligence is based on input. This can be huge sets of data, such as text or images. By looking for patterns, perhaps also assisted by a set of rules, artificial intelligence can produce outputs of value, such as suggesting solutions that have hitherto been inaccessible to humans. 

In order for artificial intelligence to function optimally, the aforementioned data sets must be optimized and structured in a way that allows them to be read and understood effectively by machines. This may require human effort. Thus, it can represent the necessary innovation in the form of human effort that will be able to trigger a patent. 

So, we are faced with two important conditions, at least one of which should be represented in a patent application based on AI-generated material: 

  1. Data is collected and prepared for AI analysis in a new and better way 
  2. The training methodology is innovative 

Does AI bring new rules of the game for patent applications? 

The Patent Office has served human innovation for centuries, and despite tweaks, the basic idea is still to reward inventors by giving them exclusive rights to their inventions for a limited period of time. None of this has changed, as machines cannot be granted a patent. 

So, the rules of the game have not changed. What has changed is how we play the game.