As a Silicon Valley patent attorney with over 20 years of experience patenting, I have seen a surge of interest in patenting AI solutions which the rage right now. Everyone wants to get in on this new technology, and they’re doing so by creating new AI applications and products as quickly as possible. But while it may seem like everyone else is getting rich off of AI, you might not be reaping any of the benefits—and that’s because you’re not patenting your AI inventions!

If you are one of those people who has been left behind by this trend, rest assured: PatentPC can help.

table of contents

  1. What is an AI invention?
  2. Why should I patent my invention?
  3. How do I patent my AI invention?
  4. Why It’s a good idea to patent AI inventions.

What is an AI invention?

Although the term AI is often used as a synonym for “machine learning”, there are several other ways in which an invention can be considered to be based on AI.
AI inventions include machines that:
● Learn from data to make decisions or perform tasks (for example, a self-driving car)
● Can act autonomously without human intervention (for example, an unmanned combat aerial vehicle) and/or
● Are able to learn new things from experience (for example, a chatbot).

Examples of AI Patents and Patent Applications

Due to the popularity of deep learning solutions, there are numerous patents and patent applications involving AI. Examples for deep learning patent applications include:

  1. IBM’s “Method for training a deep neural network” (Patent No. US20160309719A1) – This patent relates to a method for training a deep neural network using a combination of supervised and unsupervised learning.
  2. Google’s “System and method for deep neural network training” (Patent No. US9647526B2) – This patent relates to a system and method for training deep neural networks using a combination of supervised and unsupervised learning.
  3. Microsoft’s “System and method for training deep learning models” (Patent No. US20170219336A1) – This patent relates to a system and method for training deep learning models using a combination of supervised and unsupervised learning.
  4. NVIDIA’s “Method and apparatus for training deep neural networks” (Patent No. US20180280816A1) – This patent relates to a method and apparatus for training deep neural networks using a combination of supervised and unsupervised learning.
  5. Facebook’s “System and method for training deep learning models” (Patent No. US20180177610A1) – This patent relates to a system and method for training deep learning models using a combination of supervised and unsupervised learning.

Examples of self-driving applications include:

  1. Waymo’s “Method and system for autonomous vehicle control” (Patent No. US9170129B2) – This patent relates to a method and system for controlling autonomous vehicles using a combination of sensors and machine learning.
  2. Tesla’s “Method and system for autonomous vehicle control” (Patent No. US9079160B2) – This patent relates to a method and system for controlling autonomous vehicles using a combination of sensors and machine learning.
  3. Uber’s “Method and system for autonomous vehicle control” (Patent No. US20160309719A1) – This patent relates to a method and system for controlling autonomous vehicles using a combination of sensors and machine learning.
  4. General Motors’ “Method and system for autonomous vehicle control” (Patent No. US20170219336A1) – This patent relates to a method and system for controlling autonomous vehicles using a combination of sensors and machine learning.
  5. Baidu’s “Method and system for autonomous vehicle control” (Patent No. US20180280816A1) – This patent relates to a method and system for controlling autonomous vehicles using a combination of sensors and machine learning.

Why should I patent my invention?

There are many reasons to patent an invention. The most obvious is that it protects your idea, and therefore your investment. If you haven’t patented an AI invention yet but want to, then you should do so as soon as possible.
You can also get a better return on investment by licensing the technology or selling it outright. An investor who likes your idea could even buy out your company and make money from it!

can i patent my invention if I relied on open-source code?

Most AI applications rely on open-source code to some extent. Common open-source code libraries for deep learning and AI software that are widely used by researchers and developers include:

  1. TensorFlow: Developed by Google, TensorFlow is a powerful open-source library for deep learning, machine learning, and other computations. It is widely used for building deep neural networks, natural language processing, and computer vision applications.
  2. PyTorch: Developed by Facebook, PyTorch is another popular open-source library for deep learning and machine learning. It is known for its flexibility and ease of use, and is often used for natural language processing and computer vision applications.
  3. Keras: Keras is a high-level neural networks API, written in Python and capable of running on top of TensorFlow, CNTK, or Theano. It is easy to use, and often used for building deep learning models.
  4. Caffe: Developed by Berkeley AI Research, Caffe is a deep learning framework for image and video processing. It is known for its fast performance and is often used for image classification and object detection.
  5. Theano: Theano is a Python library that allows you to define, optimize, and evaluate mathematical expressions involving multi-dimensional arrays. It is widely used for deep learning and machine learning applications.
  6. Scikit-learn: It is a machine learning library for Python, built on NumPy, SciPy, and matplotlib. It’s simple, efficient and often used to perform supervised and unsupervised learning.

These libraries are widely used and well-documented, making it easy for developers to find tutorials and examples to get started with building AI and deep learning applications

It is possible to patent an invention that relies in part on open-source code, and the outcome may depend on the specific circumstances of the case. To be patentable, an invention must be novel, non-obvious, and fully and clearly described in the patent application. If an invention relies on open-source code, the patent office will consider whether the invention is novel and non-obvious in light of the open-source code.

If the invention is an improvement over the open-source code and it is novel and non-obvious, it may be patentable. However, if the invention simply uses the open-source code without adding any significant new features, it may not be considered novel and non-obvious, and thus not patentable.

Additionally, US patent laws and regulations require that the patent applicant must disclose all prior art that is material to the patentability of the invention, including open-source code, this is why it’s important to conduct a thorough prior art search before filing a patent application.

To address novelty, your patent should focus on how you formed your software to address a specific problem, and how you developed software that improved computer performance in addressing the problem you are solving. When your focus is on the problem and your solution to the problem, the fact that you used open-source code as part of your solution is akin to your use of screws and nails in forming a mechanical device. In such a case, your application would not focus on the details of how the screws/nails are formed, but how you use such screws/nails to form a solution to your problem. Thus, you show that the novelty of your invention is in the way you address the problem, not your use of open-source code.

It’s recommended to consult with a patent attorney who has experience in the field of software patents to help you navigate the process and ensure that your invention is patentable.

In summary, it is possible to patent an invention that relies in part on open-source code, but you should focus on the problem facing you and how you developed a solution that solves the problem in an efficacious manner, and the outcome may depend on the specific circumstances of the case, such as the degree of originality of the invention and whether the invention is novel, non-obvious and fully and clearly described in the patent application. It’s recommended to consult with a patent attorney who has experience in the field of software patents.

How do I patent my AI invention?

Once you’re confident that your invention is patentable, it’s time to prepar and file a patent application. An invention disclosure for an AI patent application should include the following details:

  1. Detailed description of the invention: The invention disclosure should provide a clear and detailed description of the AI technology, including its purpose, function, and how it works. It should also include any specific algorithms, models, or techniques that are used.
  2. Problem being solved: The invention disclosure should clearly describe the problem or need that the AI technology is addressing and how it solves it.
  3. Novelty and non-obviousness: The invention disclosure should explain how the AI technology is novel and non-obvious, and how it differs from any existing technology.
  4. Technical drawings or diagrams: The invention disclosure should include technical drawings, flow charts or diagrams that help to explain the invention. These can include flowcharts, block diagrams, and schematics.
  5. Examples of use cases: The invention disclosure should provide examples of how the AI technology will be used and the benefits it will provide.
  6. Data sets: The invention disclosure should include any data sets used to train or test the AI models.
  7. Prior art: The invention disclosure should include a description of any prior art (previous patents, publications, etc.) that may be relevant to the invention.
  8. Claims: The invention disclosure should include the claims, which define the specific aspects of the invention that the applicant wants to protect.

In addition, as AI inventions are software, the patent application should be written in a way to anticipate the Section 101 rejection. To show that a software invention is patentable subject matter, it should meet the criteria set forth in 35 U.S.C. § 101 of the U.S. patent laws, which states that any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, is eligible for a patent. To overcome a 101 rejection, an inventor can demonstrate that the software invention is tied to a particular machine or apparatus, or that it transforms a particular article into a different state or thing. One way is to explicitly discuss how your approach improves computer performance. Additionally, the inventor can argue that the invention provides a specific solution to a problem in a new and non-obvious way. It’s important to consult a patent attorney to help you with the process.

It’s important to note that the invention disclosure should be as detailed and specific as possible, and it should be easy to understand for someone who is not an expert in the field. It’s also recommendable to consult with a patent attorney who has experience in the field of AI to help you navigate the process and ensure that the invention disclosure is complete and accurate.

In summary, an invention disclosure for an AI patent application should include a detailed description of the invention, the problem being solved, the novelty and non-obviousness of the invention, technical drawings or diagrams, examples of use cases, data sets, prior art and claims.

Once you have the final version of the patent application, you can file the patent application with the patent office. The best way to do this is through the USPTO website and its online EFS-Web system. In order to submit an application, you will need:
● Inventor’s Oath (and an assignment to the company)
● A detailed description of how the AI invention works
● Drawings that show what the invention looks like (you can use simple sketches or photos)

How can we help?

● We can help you with all the steps of the patent process, from filing to prosecution.
● We have a team of patent attorneys and patent agents who are experts in this field.
● Our website has a powerful search tool that can help you find relevant patents and check if your idea is novel, non-obvious, and inventive enough for a patent application.
● Our website also has an easy-to-use tool for checking whether your invention is eligible for patent protection based on current laws in different countries around the world. This tool will provide you with feedback on how strong your case is regarding novelty, non-obviousness, and inventiveness so that when you decide to apply for a patent or not (and if so), it’ll be based on solid facts rather than guesswork alone!

Why It’s a good idea to patent AI inventions.

What is a patent?

A patent is a right granted to an inventor that allows them to prevent others from making, using, selling or importing their invention for up to 20 years.

If someone were to copy your patented invention without permission, they could be liable for damages. When you file for a patent, it’s not guaranteed that you’ll receive one; however, it greatly increases your chances of doing so because it helps the Patent Office decide whether your invention qualifies as new and non-obvious (which are requirements for getting one).

Conclusion

We hope that this guide has helped you understand why it’s important to patent AI inventions. If you have any questions about patenting your AI invention or would like more information, please don’t hesitate to reach out and get in touch with us through PatentPC!