The debate around AI in patent practice often focuses on whether AI can draft a patent specification. However, that that is not the question we should be asking.

The better question is: Do you know what the criteria are for a good patent specification?

Because if you cannot answer that question, you are unlikely to know whether the AI has done a good job.

Patent drafting is not a writing exercise. A specification is the combination of legal judgement, technical understanding, prosecution experience, commercial awareness, and strategic thinking.

Clients see the finished application. They do not see the hundreds of professional judgements that shape it: from identifying the inventive concept, to selecting claim scope, to anticipating future objections, litigation risks, licensing opportunities, and commercial objectives.

A well-trained patent attorney is, in many ways, a product of thousands of professional decisions. Every Office action, validity challenge, litigation outcome, client input, and drafting decision contributes to an understanding of what makes a patent not only grantable, but valuable, enforceable, and commercially relevant.

AI doesn't replace strategic decision-making

AI can generate text. It can even generate convincing patent applications.

But patent drafting is not about generating words. It is about making decisions and creating a legally and technologically solid document.

  • What is the inventive concept?
  • What should be claimed?
  • Which fallback positions should be included?
  • What language or parts of disclosure may create problems years later?

These are strategic decisions that determine the long-term value of a patent.

Experience contributes something even more valuable: the ability to identify what Donald Rumsfeld famously described as "unknown unknowns", the things we don't know that we don't know. In patent drafting, these are often the questions we never thought to ask, the risks we never anticipated, and the opportunities hidden between the lines of an invention disclosure.

Those of us who remember the pre-online era may also remember spending hours at the patent office with drawers full of classification cards. It was slow, sometimes frustrating, and nobody wants to go back. Yet there was one unexpected benefit: while looking for one document, you often stumbled across another that changed your entire understanding of the prior art.

AI, like modern search tools, is exceptionally good at helping us find what we ask it to find. Experience, however, is what tells us whether we are asking the right questions and often helps us recognise what we never thought to look for in the first place.

This is not an argument against AI. Quite the opposite.

AI can accelerate drafting, suggest alternatives, identify inconsistencies, and reduce time spent on routine tasks. The opportunity is not to resist AI, but to use it intelligently.

It is becoming one of the most powerful tools ever introduced into patent practice.

The practitioners who will benefit most are not those who simply ask AI to draft a specification. They are those who know enough to correctly guide the AI, challenge it, recognise what is missing, and refine its output into a strategically strong patent.

AI does not replace expertise. It amplifies it.

The challenge is no longer whether AI can draft a patent specification. The challenge is whether the person using it has the judgement to recognise a good one and the experience to see what is missing.

Ultimately, the value of AI-generated patent drafting will not be measured by how quickly a specification is produced or how convincing it appears. It will be measured by whether the resulting patent successfully protects innovation when it matters most.

FICPI’s view

Membership of FICPI's global federation makes IP attorneys more effective through its dual focus on legal and professional excellence.

FICPI is working to make sure that the independent IP attorney profession remains fit for purpose by adapting to the present and the future, by adapting to new challenges such as AI, continuing to bring younger members into the federation, and continuing to help drive and shape evolving IP laws and systems.

Next steps

 

Dr. Maya Shmailov is Co-Chair of FICPI's Practice Management Committee 2 (PMC 2), focusing on technology, artificial intelligence, and their impact on the IP profession. Her current work explores the intersection of innovation, law, and emerging technologies, with a particular interest in how AI can enhance patent practice while preserving the professional judgment that underpins high-quality patent drafting.