This is about how generative AI could affect our writing:
It seems that generative AI is good at creating clear and plain language, it can put difficult matters into simple terms. This ability inspires both anticipation and dread: anticipation that AI will bring relief from the grinding labour of writing well, dread that our own skills will turn out to be so inferior that they will no longer be needed.
A more productive reaction may be to feel a sense of competition. We need to process problems through the brain’s speech area to analyse and understand them. The more we put thoughts into our own words, the better we understand them. That being so, one might as well take the challenge and see who can write better sentences – AI or flesh and blood?
The clarity and conciseness of the language that AI generates may soon become the norm. The more that happens, the less readers will be prepared to struggle with language that is unnecessarily complicated. Therefore, there may be more reason than ever before to question the habits and traditions of legal writing, which often create barriers between the writer and the reader.
When you have a detailed and difficult thing to say, you should break it into several sentences – that is a truism of practical writing. However, the opposite approach is taken by some (mostly Continental European) courts. Some good examples are reference questions by national courts and the resulting conclusions by the ECJ. Here is a LINK to the conclusion of the judgement of 27 April 2023, C‑104/22 (see bottom of linked page). The language of that conclusion may be impressive in an athletic sense, but most readers equipped with a human working memory probably struggle to understand it. Anyone who has little time and patience and is required to read and understand such a conclusion might be tempted to prompt the chatbot of choice to deliver a “translated” version. This is undesirable for an author who truly wants to get the message across to as many people as the matter in question allows. If nothing but a machine adaption reaches the addressee, nuance and detail are lost on the way. To accept the growing competition by AI would mean writing in a way that the reader can easily understand.
In that sense, here is a modest attempt at conveying the ECJ’s conclusion in a fashion that is more friendly to the reader:
The reference question is about jurisdiction pursuant to Art. 125(5) of the European Union Trademark Regulation. The question has come up in a case, where the proprietor of an EU trademark considers that a sign used by a third party in online advertisements and online sales offers infringes that trademark. The proprietor has brought an action against the third party in a Member State in which consumers and traders targeted by those offers or advertisements for sale are located. The third party does not expressly and unambiguously list that Member State among the territories to which the goods in question might be supplied. But the third party has used the sign by means of paid referencing on a search engine website which uses a national top-level domain name of that Member State. In such a case, that Member State is a legal venue for an action for trademark infringement pursuant to Art. 125(5) of the European Trade Mark Regulation. By contrast, that is not the case if the third party has only used the trademark as a meta tag for images of its goods on an online photo-sharing service under a generic top-level domain.
Of course, there are many other and probably better ways to disentangle the original version. But if that first attempt has produced something that is a bit easier to read, it has been worth the effort. Making such an effort will benefit consumers of legal writing and may keep us on the ball in our competition with AI at the same time.
 Regulation (EU) 2017/1001 of the European Parliament and of the Council of 14 June 2017 on the European Union trade mark