Artificial intelligence (AI) technologies are reshaping industries, but not every AI-related innovation qualifies for patent protection. Under 35 U.S.C. §103, an invention must be non-obvious to be patentable. Simply put, a non-obvious invention is one that a skilled professional would not view as a predictable step forward based on existing knowledge. For AI inventions, this often means showing that the use of AI is not just a substitution for an existing process but delivers measurable improvements.
At Jones Intellectual Property, we help inventors, startups, and businesses secure patents that withstand legal scrutiny. Our firm, led by Michael Jones, combines extensive technological knowledge with personalized and affordable intellectual property services. We assist clients with preparing strong patent applications, responding to examiner rejections, and framing their AI innovations to highlight inventive steps.
Non-Obviousness (§103) and AI Inventions
To qualify for a patent, an invention must be eligible for protection, novel, non-obvious, and reproducible. Under 35 U.S.C. §103, non-obviousness means that even if an invention is new, it does not qualify for a patent if it is an obvious next step from existing technologies. In other words, only non-obvious innovations are patentable.
Non-obviousness can present an obstacle for AI patent applications. Many AI systems use machine learning models to replace traditional or manual methods of performing tasks such as data analysis, image recognition, medical diagnostics, or document review. Patent examiners at the United States Patent and Trademark Office (USPTO) often conclude that such substitutions lack a genuine inventive step, making them obvious.
Understanding Obviousness in Patent Law
When considering obviousness in patent law, USPTO examiners make decisions based on what a “person having ordinary skill in the art” (abbreviated as PHOSITA) would think to do with existing technologies. PHOSITA serves as a benchmark for judging whether an invention requires more than routine insight. Examiners use PHOSITA to ask: “Would someone familiar with current techniques reasonably develop this invention without additional creativity?”
Prior Art
To determine what current techniques include for the PHOSITA inquiry, USPTO examiners consult industry standards along with “prior art,” which may include:
- Earlier patents and published patent applications;
- Scientific or technical journal articles;
- Conference papers and presentations;
- Industry manuals, handbooks, or textbooks;
- Commercial products or services already in use; and
- Public demonstrations, sales, or disclosures of technology.
If the invention is an obvious improvement to any of these already disclosed technologies, it is not patentable.
Patent Obviousness Rejections
Common scenarios that lead USPTO examiners to reject patent applications as obvious include:
- Combining known elements in a predictable way—adding a touchscreen to a calculator after touchscreens became standard in phones and tablets;
- Substituting one material for another—using plastic instead of metal for a container when the choice of material would not affect performance;
- Making routine size or shape modifications—scaling a product up or down, such as enlarging a smartphone screen, without introducing new functionality; and
- Adding features that professionals already expect—including Wi-Fi or Bluetooth in a common device when the market widely anticipates that upgrade.
Examiners often reject as obvious AI systems that:
- Replace a manual step in a process with AI, like using AI to sort incoming emails;
- Adjust obvious parameters of a machine learning model, such as the dataset size; or
- Apply off-the-shelf AI tools to routine tasks, like fraud detection or spam filtering.
Unless these innovations lead to surprising results, USPTO examiners likely consider them obvious.
Challenges of Obviousness in AI Inventions
AI patents face unique scrutiny because examiners often treat them as routine substitutions—simple replacements of parts of other inventions that do not create real innovation. USPTO examiners frequently consider inventions obvious when:
- AI replaces an existing process without changing the outcome;
- Algorithms handle standard tasks and do not produce results that would surprise a skilled professional; and
- Claims fail to show measurable technical advantages, such as improving accuracy by a specific percentage, reducing processing time, or enabling scalability that prior systems could not achieve.
For example, applications that attempt to patent “using AI for image classification” often fail unless they show how the approach improves upon existing methods. Without highlighting results such as faster recognition of what the image includes or reduced computing demands, the invention looks like an obvious extension of known techniques.
Best Practices for Drafting AI Patent Applications
To overcome patent obviousness rejections, inventors must carefully frame their claims, including:
- Emphasizing inventive steps,
- Highlighting unexpected results,
- Providing comparative evidence, and
- Explaining technical advantages.
By grounding the application in specific, non-obvious advancements, innovators strengthen their case before the USPTO.
Takeaways for Innovators
Meeting the requirements of 35 U.S.C. §103 requires showing non-obviousness in patent law—that the invention delivers more than a predictable substitution or minor modification. Innovators should remember that:
- An invention that represents a routine or obvious extension of existing methods is not patentable;
- AI patent applications should stress the inventive step and highlight unexpected benefits with measurable data; and
- Comparisons to prior methods strengthen arguments against obviousness and clarify why the invention goes beyond predictable outcomes.
Innovators should frame their AI inventions as solving problems in ways that industry professionals would not reasonably expect.
Frequently Asked Questions (FAQs)
How Does Obviousness Apply to AI Inventions?
Obviousness applies when an AI invention looks predictable to someone skilled in the field. Examiners review prior art and industry knowledge to decide whether the invention flows naturally from existing information.
Why Do Examiners Often Reject AI Patent Applications for Obviousness?
Examiners often view AI inventions as routine substitutions, such as applying an algorithm to automate a manual process. Without unexpected results or clear technical improvements, these applications fall short under 35 U.S.C. §103.
What Strategies Help Innovators Overcome an Obviousness Rejection in AI Patents?
Applicants should emphasize inventive steps, highlight unexpected advantages with concrete data, and provide comparative evidence showing how the invention performs differently from existing methods.
What Should Innovators Focus On When Drafting AI Patent Applications?
Innovators should frame their inventions to show how they solve problems in ways that skilled professionals would not expect. They should highlight unique system designs, surprising improvements, and measurable advantages over prior approaches.
Moving Forward with AI Patent Protection
At Jones Intellectual Property, we guide inventors and businesses through the challenges of avoiding obvious AI patent applications, helping them demonstrate inventive steps that stand out under 35 U.S.C. §103. If you are developing AI technologies, our firm can provide the insight and legal strategies you need to protect your innovation. Contact us today to discuss how we can safeguard your ideas and ensure your application meets patentability standards.
