From ChatGPT to smart indoor farming to speech recognition technology that decodes baby cries, artificial intelligence (AI) is an undisputed game-changer. According to IBM, the global AI adoption rate sits at 35%, with an additional 42% of companies actively exploring AI. AI is transforming many aspects of everyday life, from manufacturing and the supply chain to agriculture to healthcare and government.
From 2002 to 2018, annual AI patent applications increased by more than 100% to more than 60,000 annually. This corresponds to the share of all patent applications that contain AI from 9% to nearly 16%. By 2020, USPTO had received about 80,000 patent applications for AI software and approved 77% of them. Whether it is a machine learning algorithm, which makes up 40% of AI IP cases, natural language processing, or computer vision, AI IP rights stand poised to shake up the business landscape.
This guide will provide a detailed breakdown of patenting artificial intelligence and why intellectual property rights are increasingly important in AI development.
Understanding The Basics: Patenting Artificial Intelligence
Whether you are an inventor or represent a company, you may be wondering what artificial intelligence is even patentable? Roughly 100,000 organizations worldwide own AI patents, with a large volume of them coming from global tech giants such as Samsung and IBM.
Types of Patents Available for AI Technology
While AI feels complex, its applications often boil down to a few key areas. First, machine learning makes up the majority of AI techniques and has grown roughly 28% each year from 2013 to 2016. Under “machine learning,” deep learning and neural networks give way to machines that truly “learn” upon their own work. Algorithms learn, iterate, and improve upon their outputs through modeling and reinforcement.
Product recommendations are a prime example of machine learning; based on your shopping and purchasing behaviors online, recommendation engines can suggest products you may like. Then, based on your response to those recommendations, the algorithm can narrow your preferences even further, creating more fitting suggestions.
Second, applications such as computer vision, facial recognition, natural language processing, and speech processing make up a bulk of artificial intelligence uses. Facial recognition technology is often used in social media filters and biometrics are also used now for increased border security and safer travel.
The biggest sectors with AI patents are telecommunications, transportation, life, and medical sciences.
Key Components of a Successful AI Patent Application
What constitutes a successful AI patent application? If you have ever applied for any kind of IP rights, you know it can be a time-consuming and tricky process to navigate. There are a few key components of a successful AI patent application.
According to the United States Patent and Trademark Office (USPTO), a patentable technology must be a “new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement.” Most successful AI filings can demonstrate the application of the AI in a specific practice.
Patentable Subject Matter in Your AI Software
Patentable software must be more than simply layering AI over a known practice or system. AI-based technology must be new, non-obvious, and beyond a simple improvement of an existing application. According to the United States Patent and Trademark Office (USPTO), a patentable technology must be a “new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement.” Most successful AI filings can demonstrate the application of the AI in a specific practice.
AI systems must be new, non-obvious, and pertain to something that is patentable. The patent application must also accurately and thoroughly describe the AI, and it must be an original work. AI must solve a real-world or computer problem. This could involve solving the problem of tracking heart palpitations in cardiac patients or even improving a computer model to perform faster and more accurate outputs.
The USPTO has identified a small group of example patentable technologies implementing AI. Patent protection may be available for software implementing AI, including:
- Novel pre-processing of model training data. This may include generating unique datasets prior to training an AI model.
- Novel model training processes. This may include which algorithms are used in a process or improvements to a machine learning model or neural network algorithm.
- Novel use of trained models. This may include implementing AI to control specific machines or to provide unique results from known products or methods.
- Improvement to the underlying hardware implementing a trained AI model or algorithm or the production of a tangible result or object.
These examples are not a complete list of potentially patentable subject matter, and in many cases, AI-based software may include several of these novel features. Both the United States Patent and Trademark Office and European Patent Office are constantly updating regulations due to the boom in AI patent filings. Legal rulings and qualifying innovations might change as they attempt to keep up with the latest practices and innovations. As more and more companies grab up AI-based patent territory, it is increasingly more important for businesses to submit patent applications to protect their technology.
The Role of AI Patents in Creating a Competitive Advantage
Protecting AI Inventions from Unauthorized Use
One of the main reasons companies want to patent anything is to protect it from unauthorized use. This is especially true for artificial intelligence patents due to the significant competitive edge they can provide. Not only can companies market their brand as innovative, but they can also use AI to outpace competitors and win more business.
Issued patents are valuable assets that add value to your company, create licensing revenue, and position you for cornering entire markets. A robust AI portfolio can also demonstrate security and growth to potential investors. Companies implementing cutting-edge AI should focus on protecting their intellectual property through patents. Similarly, even companies that are “fast followers” have ample opportunity to carve out their market by using improvement patents to cement their value.
AI IP is a sound investment for startups and large corporations alike. Creating a valuable IP portfolio allows companies to protect current inventions while capturing white space in the market. This might help companies raise venture capital as they can lean on a differentiator of trailblazing and discovery.
A robust patent portfolio can transform a company from merely selling products and services today into a company that has protected its bottom line for years to come. Issued patents live for 20 years from their filing dates, which means products and services with patent protection have a clearly defined monopoly in their market.
The Patenting Process for AI Technologies
Both the United States Patent and Trademark Office and European Patent Office are constantly updating regulations due to the boom in AI patent filings. Legal rulings and qualifying innovations might change as they attempt to keep up with the latest practices and innovations. Thus, the process is similar to filing for any sort of intellectual property right.
1. Identify Patentable AI Innovations
Speak with a qualified patent attorney at the Rapacke Law Group to identify patentable subject matter in your AI-based software or within your own AI models and training processes. A Rapacke Law Group will help you research and prepare your patent application to meet the requirements of the USPTO, such as 35 U.S.C. 101, 102, 103, and 112.
2. Conduct a Thorough Prior Art Search
Your technology needs to be new and non-obvious to achieve patentability.. This requires conducting a thorough patent search to ensure you’re not infringing on an existing patent and gathering research on successful similar patents. This helps build the strongest case for protecting your innovation.
3. Prepare and File the Patent Application
Once you have determined your AI is, in fact, novel and nonobvious, it’s time to file your patent. First, clearly identify the novel, inventive aspect of your AI. Thoroughly describe all systems, processes, and components involved, and avoid generic language or simply re-writing your AI’s functionality as replacing a human.
Overcoming Potential Rejections and Objections
Not every patent application is successful based on the initial submission. Your patent application may either be accepted, rejected, or objected to. Rejection happens when the patent content is “unallowable,” but an objection happens when something is simply incorrect, such as a typographical error. Patent Examiners make a rejection determination when the issue involves the merits of the claim, and you can ask for a review by the Patent Trial and Appeal Board (PTAB). Typically, rejections relate to the law, and objections relate to the rules.
Neither rejections nor objections spell the end of the road for your patent application. Objections can typically be overcome by amending the content of your application. Rejections can typically be overcome by amending the content of your application and submitting tailored legal arguments for the patentability of your invention. This process, called “prosecution,” is a necessary step to ensure that the USPTO issues quality patents that have been vetted against existing technology. A Rapacke Law Group patent attorney can assist you in responding to rejections and objections.
Speak With an Attorney Experienced in AI Patents
The Rapacke Law Group is an intellectual property law firm specializing in AI. With the combined knowledge of expert staff, including former USPTO Primary Examiners and AI developers themselves, we’re ready to help you secure your intellectual property and leverage your competitive advantage to the fullest extent.
Schedule a free strategy call to learn more about leveraging AI and subsequent IP rights to your advantage.
