Artificial intelligence and trade mark assessment

Artificial intelligence has an unparalleled potential for facilitating IP administration processes, in particular in the context of examining trade mark applications (e.g. descriptiveness or immorality of terms) and assessing prior marks in opposition and infringement proceedings (e.g. assessing the likelihood of confusion with another sign). Several stakeholders have developed AI-based algorithms that are claimed to enhance the productivity of trade mark professionals by carrying out, without human input, (parts of) the legal tests required to register a trade mark, oppose it or claim an infringement thereof.

The goal of this chapter is to assess the functionality of the AI tools currently used and to highlight the possible limitations of AI tools to carry out autonomously the legal tests enshrined in trade mark law. In fact, many of these tests are rather subjective and highly depend on the facts of the case, such as an assessment of the distinctive character of a mark, whether the relevant public is likely to be confused or whether a third party has taken unfair advantage of a mark. We use doctrinal research methods and interview data with 14 stakeholders in the field, including algorithm developers, trade mark examiners, judges, in-house trade mark counsels and attorneys from different jurisdictions. We find that so far, AI tools are unable to reflect the nuances of the subjective legal tests in trade mark law. Even in the near future, we argue that AI tools are likely to carry out merely parts of the legal tests and present information that a human will have to assess, taking prior doctrine and the circumstances of the case into account.