Seven AI tools for trademarks
- I is going to drive the development of new tools in the trademark field
- It is likely to be the subjective elements of trademark law where AI has the most potential to transform trademark work
AI is going to drive the development of new tools in the trademark field, but before considering the effect they will have, it is worth considering what kinds of tool they will be. AI is developing fast, and the term spans a wide range of computing techniques and applications – some of which are more obviously relevant to trademark work. Some of the most interesting are explored below.
The ability of machines to learn from new data, without being programmed by a human, is vital for trademark classification and searching, where the volumes of data are growing fast. Such technology has been available in some form for more than 20 years and is widely used by both IP offices and private providers. It copes well with word marks and simple shapes but struggles with more complex shapes.
Deep learning is a form of machine learning that represents the world as a hierarchy of concepts and uses huge computing power to make much finer decisions than older forms of machine learning. It has the potential to analyse more complex trademark data, in greater volumes, in different categories and from various sources, and could be used to assess complex shapes and other non-traditional marks. There is currently a lot of money being invested in deep learning, much of it being spent on the considerable efforts required to sort and input the volume of data required.
Unlike Boolean logic, where variables may be only 0 or 1, fuzzy logic accommodates facts that can be anywhere between totally true and totally false. Therefore, it has great potential to be applied to the evaluation of trademark concepts, such as the similarity between two different signs or between different goods and services.
Another technology that is attracting a lot of investment, neural networks are modelled on the human brain and can solve tasks that are too difficult for conventional computer systems, for example, assessing a wide range of evidence in different formats or detecting relevant images in a messy environment. They drive deep learning and could be useful in complex trademark cases, in anti-counterfeiting campaigns or in assembling evidence of use and non-use.
Computer vision is the field of studying digital images or videos. It is already well developed in the consumer field, but its application to trademarks is just beginning to be explored, for example, in identifying counterfeit goods online.
Natural language processing
Natural language processing is the field of AI that deals with computers processing and analysing human languages with all their complex grammar rules and exceptions. Natural language processing can be used to turn information written in human words and languages (eg, descriptions of goods or services) into something that computers can understand and analyse.
Predictive analytics uses data, models and other AI tools to predict outcomes with a high degree of confidence. Predictive analytics could be used to forecast the outcome of trademark applications or disputes based on the interpretation of thousands or even millions of previous cases. Taken to its logical conclusion, such technology could deliver more predictable results and do so more efficiently than trademark examiners or even judges.
Computers are often seen as poor substitutes for human judgements, but as these technologies demonstrate, it is likely to be the subjective elements of trademark law (eg, what is distinctive? When are two things similar? What does the average consumer think?) where AI has the most potential to transform trademark work.