What has Changed Since Launching Our Patent Drafting Course

AI happened

Making courses on patent law is a resource-draining, time-consuming endeavor.

The European Patent Academy was established by the European Patent Office (EPO) to educate and train a diverse audience of patent professionals, judges, public servants, business managers and academics. Since 2004 it has launched about 120 courses. That's at an average rate of 6 courses a year.

Few years ago, we had launched our courses, Patent Law for Engineers and Scientists and Patent Drafting for Beginners. Like the EPO courses, they are timeless and have served well for all these years. The course on Patent Drafting has remained the same without any changes since its launch.

After the launch of the course, something happened that has required us to revisit the Patent Drafting course in a new light.

Since its launch, AI happened.

And AI changes almost everything about the patents. It changes how patents are drafted and prosecuted. It should also change how patent drafting is taught and learned. AI's impact on patents can be seen in two areas which have an enormous impact on how patents are drafted and prosecuted.

AI assisted Patent Drafting

AI-enabled patent drafting software have changed the domain for good. Some of the objections that are routinely raised in office actions will soon be a thing of the past, as AI-enabled tools will detect and correct them in an instant. Not only do these tools automate time-consuming tasks like claim formatting, reference numbering and term consistency, they also flourish in creative tasks like generating background, description of drawing, title and abstract. These tools can also be used for generating responses to office actions.

AI assisted Prior Art Search

Google set out to organize the world's information. Google also understood that the patent system forms a critical part of the world's information. One of the training datasets on which AI models are trained, the C4 (Colossal Clean Crawled Corpus) relies extensively on patent documents as the single most represented website in the corpus is patents.google.com. USPTO Backgrounds have been used to train another dataset called "The Pile". The EPO also boasts of having the biggest search database for screening prior art. The ability to machine-translate documents from one language to another further multiplies the amount of data. In the 1990s, the EPO had special expertise in machine-translation which was the reason why Google partnered with them in their translation project. The EPO today allows you to machine-translate documents in 30-odd languages. Given the enormous amount of published data that a single patent has to be screen before its grant, AI-enabled search tools are fast emerging as the tool of choice to save time and money. Prior art searches that used to take days can now be done in few minutes. Here's what one the AI enabled platforms say about their tool:

The AI-based Novelty Checker from XLSCOUT can help you conduct a quick first pass prior art analysis and generate a novelty report in 5-7 minutes.

A patent needs to satisfy two types of requirements: internal and external. With regard to the internal requirements (such as enabling disclosure, written description, sufficiency of claims, best mode, antecedent basis etc), a good software can most certainly catch them and correct them. With regard to the external requirements (such as novelty and inventive step), the better AI-enabled search gets, the closer we will be to anticipate and address such objections. AI-enabled tools on patent drafting and prior art search may bring us close to the ideal state of anticipating all objections and creating bulletproof patents.