Introduction

Today, in the field of patents, Artificial Intelligence (AI) is emerging as a powerful ally. By supporting inventors, IP professionals, and patent attorneys in conducting searches, managing prosecution, and streamlining the drafting process; AI is giving them time to focus on more strategic aspects of innovation. In fact, the 2024 survey conducted by FICPI’s Practice Management Committee (PMC) found that among those using AI tools for work, 38% use it for patent searches, 19% (approximately) for patent drafting, and 24% use AI for analyzing prior art.

However, AI is still developing and far from being the flawless Jarvis we imagine it to be. Its outputs, while efficient, must be approached with caution, especially concerning accuracy, confidentiality, and compliance with laws.

So, if you are considering integrating AI into your patent process, here are three important things to keep in mind. 

1. How Can AI Be Useful?

AI has changed how things are prepared, reviewed, and filed. Today, you can use AI for more than one stage of your patent application, such as:  

• Conducting Patent searches – AI can narrow down the scope of your prior art search criteria and even categorize existing patents based on grounds set by you.
• Drafting Patent Applications – AI may help you provide quick drafts within minutes, reduce efforts and hours, save money, reduce clerical errors, and even suggest broader claims (providing, of course, that the AI database supports higher level tasks). 
• Generating Patent Drawings – AI can help to generate drawings of your invention, either based on images or even text, and even aid in labeling them
• Plus, AI can even help you analyze or review patent applications.

However, these benefits are only realized if you use it in the right way, such as:

• Providing a complete invention disclosure – Think of AI as a person with no prior knowledge about your invention. To allow an AI to process your novel invention,, you will need to define all parts of it (background technology, unique features, competitive advantages etc.) in detail.
• Reviewing the AI output – Even if we input all the relevant information, the AI may not necessarily give us the desired or required output. You will need to strictly review the output before finalizing your patent draft.  

 

2. How Using AI Is Risky

While AI may be beneficial, it is not foolproof and comes with certain risks: 

• Confidentiality risks: Using AI tools without strong data protection, especially cloud-based AI tools, comes with the inherent risks related to privacy and confidentiality.  The information or prompts that you are providing the AI get stored on a third-party server, which, if released by mistake or used in training an AI model, can make you lose the novelty aspect of your patent. 
• AI hallucinations: Sometimes AI tools can generate data, citations, or facts that may seem to be true, but are actually false. Such mistakes are called AI hallucinations. These mistakes, if not addressed, can be fatal during patent prosecution . 
• Quality detection issues: Today, AI tools are being used so widely that people can decide whether a text is AI-generated or not from a small sample. So even if the content is accurate, certain phrases or patterns may reveal its AI origin, which could affect perceptions of the quality or originality of your draft.

 

3. How to Choose the Right AI

Not every “AI-powered” tool understands patents. Many simply generate text, not structured applications. Thus, it’s necessary to look out for certain essential elements when selecting an AI tool for your patent process.

While most AI tools, especially for patents, are Generative AI-based on Large Language Models (LLM), a good patent AI tool is the one that:

• Maintains confidentiality – Since you are providing the information about your patent, the  AI tool you pick must have strong encryption with no data retention policy. Local AI tools as opposed to cloud-based AI tools are much better suited to this. Thus, it is necessary to know about the privacy policy, terms of use, and data handling practices of such AI tools.
• Consider the Prior Art and have access to a database – If an AI tool has access to various patent databases not only of your jurisdiction but also of other jurisdictions as well, then this will not only help you in your prior art search but also save your time in drafting patent applications. A larger database  can better process the existing prior art from  different jurisdictions. 
• Understand the legal compliances: Any AI tool can translate or format your invention into claims by simple commands. But a good AI tool should have the capability of understanding your invention in context  with the existing law and provide drafts accordingly, such as the correct format, correct size, correct number of claims, and so on.
• Can customize and is flexible – Each patent draft is unique, and may be different based on your invention or jurisdiction (where you intend to file). Thus, an AI tool that can adapt your drafts into different templates in its output is more desirable.
• Explains the Drafts created – Don’t choose an AI that just drafts instantly without showing how claims were structured or which prior art was considered. An AI tool that offers transparency by explaining how it builds claims or maps concepts not only helps you to review it more effectively, but also lets you understand your own invention more correctly.
• Provides Drawings: If an AI tool, apart from just drafting a patent, can also provide aid in drafting drawings, then it is a cherry on the cake and would save a lot of time and effort in the preparation of your patent application.
• Aligns with your budget – AI is the future, but it’s not compulsory. Given the stringent requirements for AI-based patent drafting, careful curation of the tools you use are required. Under such circumstances, make sure that the AI tool you use aligns with your budget. 

The Key Takeaway

What must be remembered is that AI is here to assist you,  not to do the entire task. Even if you select the best AI for your patent search and drafting, it does not mean that it won’t make mistakes.

When using AI, think of the inventor as a teacher, who needs to guide the AI so that it can understand and explain the inventions as we know and want to. While certain AI tools have better capabilities and understanding than others and can provide the desired results with less effort, it’s ultimately up to the inventor to decide if it’s worth spending time and resources guiding the AI and entrusting his confidential knowledge to it.

---

Interested in innovation, technology and patent protection? I have a lot of insights into how technology protection works from my years in the field, and I’ll be sharing more of them on this newsletter.

Connect with me if you are thinking about IP Protection!